Remote monitoring of powder dehumidification by speckle pattern analysis Authors: Mahsa Asghari, Eric van Bruggen, Kamal Heidary, Ali-Reza Moradi Published: 2026 ABSTRACT Non-invasive monitoring of heterogeneous powders is a required task in several food science and industrial applications. We introduce laser speckle analysis (LSA) as a remote, non-contact, non-invasive, cost-effective, and speedy approach for tracking powder dehumidification. LSA is based on the statistical analysis of dynamic speckle patterns that are generated from scattering samples upon laser light illumination. In this work, LSA is applied to monitor dehumidification dynamics in cornstarch powder, chosen as a representative hygroscopic food system. Cornstarch is selected due to its industrial relevance as one of the most widely used hygroscopic food powders, where moisture content strongly governs flowability, caking, and shelf life. In addition, its highly diffusive structure provides a suitable testbed for LSA experiments. As drying progresses, the speckle-derived metrics exhibit specific trends that indicate a gradual reduction of internal dynamics. The optical observations are validated by gravimetric measurements and 3D profilometry. The results present a two-stage drying profile, characterized by an initial phase of rapid mass loss followed by a slower approach toward equilibrium. Collectively, the LSA results, in agreement with the validating experiments, reveal the formation and stabilization of surface micro-fractures over time. The technique effectively captures both dynamic and structural transitions, offering a practical solution for process monitoring and quality control in industrial settings, and has the potential to serve as a bench-top analysis device. Introduction Powders are widely used in food, pharmaceutical, chemical, and material industries. Their properties depend strongly on moisture content and water activity. Even small variations in moisture can change their flowability and cohesion, promote caking and agglomeration, and reduce stability and shelf life. Because of these critical dependencies, monitoring of the drying and humidity is essential to secure product quality and maintain efficient processing1,2. Different techniques have been developed to measure the moisture in powders. Gravimetric and thermogravimetric methods are widely used and standardized. However, these methods are destructive, slow, and may overestimate water. Optical approaches, such as near-infrared (NIR) spectroscopy, allow rapid, non-contact, and non-destructive measurements, but they rely on calibration models and are sensitive to sample packing and heterogeneity. Contact sensors, while holding more precision, can disturb the powder bed and suffer from long-term drift. These drawbacks highlight the need for complementary methods that can directly track micro-dynamics during drying through a non-perturbing mechanism3,4. This paper represents a non-contact, remote, non-invasive, and non-destructive alternative based on the analysis of the “speckle patterns”, the intensity distributions that are formed upon back-scattering of laser light from the powder. The formation of a speckle pattern originates from the reflection of a coherent light from a rough surface or its transmission through a diffusive environment5,6. The speckle pattern includes several bright grains surrounded by a network of dark regions. The intensity distribution is, indeed, the interference of these randomly diverging light waves, and the bright and dark regions correspond to the constructive and destructive interference events, respectively. In everyday life, the speckle patterns can be formed through interaction of laser light with various materials, such as paper, white paint, unpolished metal, translucent materials, and colloidal fluidic mixtures. While speckle patterns may initially appear as noise, they contain valuable information about the surface or the medium from which the light is reflected or transmitted. The embedded information is comprehensive but mixed-up, and requires spatio-temporal statistical analysis to be extracted. The statistical analysis has led to speckle based detection, measurement, and imaging methodologies with significant applications in multiple disciplines. Dynamic laser speckle analysis (LSA) relies on statistical processing of temporal variations in the speckle patterns of a dynamic phenomenon7. Therefore, it requires the acquisition of a series of speckle patterns over a specific period of time, defined by the dynamic time scale of the phenomenon. This method serves as a non-invasive and non-destructive method to investigate variations in diverse samples. One of its primary strengths lies in its straightforward and cost-effective experimental setup. Numerous applications spanning a range of fields, with particular emphasis on the medical and industrial sectors, have harnessed the potential of this approach. For example, we can refer to our previous researches, including the detection of electric current flowing in wires8, monitoring of bone scaffold activity9, detection of intralayer alignment in multicomponent lipids10, control and analysis of microscopic collective motion11, characterization of PVK:rGO composite based memristor device12, investigation of red blood cell morphology under air pollution13, and investigation of paper crumpling14. Several other applications have also been reported, such as optical multimode fiber-based pipe leakage sensing15, dynamic magnetic field sensing16, Gaussian beam size measurement17, evaluating blue light impact on reconstructed human epidermis18, and analysis of turbid liquids19, to name only a few. Despite the broad range of applications of LSA in biological, medical, and porous systems, its application to hygroscopic food powders remains relatively limited. In particular, the use of LSA for monitoring dehumidification dynamics in such materials has not been extensively explored. This highlights the need to investigate the applicability of speckle-based approaches in food powder systems, where moisture-driven dynamics plays a critical role. We apply the LSA methodology to the dehumidification of cornstarch powder as the model specimen. Cornstarch is selected due to its high industrial relevance and well-characterized behavior as a hygroscopic material. Its moisture content strongly affects critical powder properties such as flowability, caking, and storage stability, making it a practical and sensitive test case for drying studies20–22. Additionally, its fine particle size and highly diffusive optical characteristics render it an ideal scattering medium for speckle-based techniques, allowing robust detection of internal microdynamic changes during evaporation. To characterize the drying dynamics, we compute a suite of activity descriptors, including the time history of speckle patterns (THSP), co-occurrence matrix (COM), inertia moment (IM), absolute value of differences (AVD), motion history image (MHI), and temporal intensity autocorrelation7,9,23,24. To validate our results, we also perform gravimetric mass measurements to determine drying rates and employ a non-contact 3D optical profilometer to monitor the surface roughness of the samples. In Section 2 we provide the experimental procedure, the numerical processing details, and the sample preparation considerations. The results are presented in Section 3, and the paper is concluded in Section 4. Material and methods Sample preparation The sample used in this study is unmodified waxy cornstarch powder (Sigma-Aldrich), selected for its well-characterized hygroscopic properties and microstructural homogeneity. A mass of approximately 6 grams is carefully weighed and transferred into a shallow open aluminum container. The powder is loosely compressed using a rotating motion to create a smooth and level surface, ensuring uniform light scattering conditions across the sample area. The samples are equilibrated in a controlled climatic chamber using a bench-top incubator (Memmert GmbH, Germany) at a constant temperature of 21\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {^{\circ }C}$$\end{document} and predefined relative humidity levels (90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}, 70\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}, and 50\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}). The relative humidities are set using the incubator’s internal humidity control system. The equilibration time is at least 16 hours (typically overnight), ensuring sufficient moisture equilibration within the powder samples. Throughout the incubation time and the experiments, the recorded relative humidity and temperature are carefully rechecked. In addition, the ambient temperature and relative humidity are independently monitored using a thermometer and a digital hygrometer (Inkbird IHC-200), respectively. The preconditioning step allows the samples to reach moisture equilibrium and minimizes variability arising from prior storage conditions or ambient environmental fluctuations before the dehumidification process. All samples are prepared under identical environmental conditions using the same setup and protocol toward internal consistency of the experiments, and reliability of the reported comparative results. Experimental setup Fig. 1(a) The experimental setup to record dynamic speckle patterns. The sample is the cornstarch powder subjected to the dehumidification process. (b) The stacked speckle patterns organization for the analysis process. (a) The experimental setup to record dynamic speckle patterns. The sample is the cornstarch powder subjected to the dehumidification process. (b) The stacked speckle patterns organization for the analysis process. The experimental setup is schematically illustrated in Fig. 1(a). A He-Ne laser (Melles Griot, 632 nm, 5 mW) is used as the coherent light source. The laser beam is spatially cleaned using a standard filtering unit composed of a 10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} microscope objective and a 20 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu$$\end{document}m pinhole, ensuring removal of high-frequency spatial noise and producing a clean Gaussian beam profile. The filtered beam is then collimated using a lens with a focal length of 70 mm. After collimation, the laser beam is directed onto the sample surface using a mirror. The back-scattered light from the sample is collected at an angle of approximately 20\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{\circ }$$\end{document} with respect to the incident beam. A collecting lens with a focal length of 50 mm collects the scattered light and transfers it to a CMOS camera (Thorlabs, Zelux CS165MU), which results in the formation of a speckle pattern. Samples are transferred from the incubator to the speckle measurement setup to monitor the drying dynamics of the powder samples. To track the activity of the cornstarch powder samples, speckle patterns are recorded as continuous video sequences at a frame rate of 10 fps using 8-bit acquisition with a fixed exposure time of 0.5 ms and a gain setting of 0 dB. At time points \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_1$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_2$$\end{document},..., \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_M$$\end{document} after starting the experiment, a continuous recording is performed, from which a subset of 200 consecutive speckle patterns (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t_1$$\end{document} to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t_N$$\end{document}) is extracted for analysis. All statistical parameters (THSP, COM, IM, AVD, and MHI) are computed using these fixed-size frame sequences (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$N = 200$$\end{document}), corresponding to an acquisition duration of 20 s. The same acquisition conditions are maintained for all measurements and across all relative humidity conditions. Fig. 1(b) shows the organization of the recorded stacks of speckle patterns. Numerical statistical processing of speckle patterns Dynamic speckle pattern analysis is a non-invasive technique for monitoring the behavior of active and dynamic materials. It enables the extraction of useful information about the physical and structural changes in a sample over time. Dynamic activity occurs when the sample changes its properties due to the local movements or configuration changes of the particles that constitute it, the scattering center’s movement, variations of the refractive index distribution, etc. These phenomena influence the light-matter interaction in terms of absorption, reflection, and scattering, and drive the formation of dynamic speckle patterns. The speckle patterns for monitoring of powder dehumidification will fluctuate in time due to the progression of evaporation and its impact on the microstructure of the sample. As water evaporates from the surface and interior of the powder bed, capillary forces and moisture gradients may induce displacement or rearrangement of particles, i.e. light scatterers. By continuously recording speckle patterns during drying and analyzing their temporal variations, one can monitor the dynamic behavior of the sample in real time. Also, the stability of the laser’s wavelength and intensity ensures that the observed speckle dynamics are solely attributed to changes within the sample itself, rather than fluctuations in the light source. To extract the information from the speckle patterns, several statistical parameters, which are listed and described in the following, are calculated. Time history of speckle patterns (THSP) Time or temporal history of speckle patterns (THSP) is a 2D matrix that encapsulates the temporal progression of consecutive SPs. THSP is formed by random selection of M points throughout each pattern of the N successively recorded SPs, and arranging such M \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} 1 columns next to each other. Therefore, the new composite image, THSP, will have M \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} N size; the rows (M) of this matrix symbolize the set of points, and the columns (N) represent their intensity state at each sampled moment, hence, signifying the temporal progression and the total number of patterns. THSP is utilized to characterize the activity in samples, with activity manifesting as intensity changes in the horizontal direction. Thus, greater intensity variations in the THSP horizontal lines correspond to samples with higher activities. Moreover, the calculation of the THSP matrix acts as the basis for obtaining additional analysis parameters, such as the co-occurrence matrix and inertia moment. Co-occurrence matrix (COM) The co-occurrence matrix (COM) represents the probability of transitioning between intensity values in two adjacent pixels of a THSP matrix. This transition histogram of intensities illustrates the sample’s activity as the dispersion of non-zero values beyond the main diagonal. The COM is defined as follows:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \mathrm{{COM}}(i,j)=\sum _{m=1}^{M}\sum _{n=1}^{N-1} {\left\{ \begin{array}{ll} 1, & \hbox { if} ~~ \mathrm{{THSP}}\ (m,n) = i \\ & \text {and}~~\mathrm{{THSP}}(m,n+1) = j,\\ 0, & \text {otherwise.} \end{array}\right. } \end{aligned}$$\end{document}Here, i and j denote the intensity of two adjacent pixels. For an active sample, the intensity values evolve over time, the non-zero elements near the main diagonal increase, and the matrix takes on a cloud-like appearance. Conversely, for a low-activity sample, the matrix values are concentrated around the main diagonal, signifying more diverse and varied intensity transitions25. The intensity values used in COM are integers, as, according to the above definition, it is constructed by counting the frequency of co-occurrences of specific intensity pairs within a defined spatial relationship across the pattern. The size of the COM matrix is not determined by the dimensions of the pattern (M \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} N), but rather by the range of intensity values present in the pattern25. Specifically, if the image has L distinct intensity levels, e.g., 256 for an 8-bit digital recording device, the COM will be a L \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times$$\end{document} L matrix. Each element in this matrix represents the number of times a pair of intensities occurs with a specified spatial relationship in the pattern. Inertia moment (IM) Inertia moment (IM) quantifies the dispersion of values around the COM principal diagonal. In analogy with mass moment of inertia, IM is defined as the accumulated COM matrix values that are multiplied by the square distance to the original diagonal and normalized26:2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \mathrm{{IM}} = \sum _{i}^{}\sum _{j}^{} \frac{\mathrm{{COM}}(i,j)}{\sum _{m}\mathrm{{COM}}(i,m)}|i-j|^2. \end{aligned}$$\end{document}By definition, diagonal elements of COM do not contribute to IM, as their distance \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$|i-j|$$\end{document} is zero. In contrast, off-diagonal elements, particularly those farther from the diagonal, contribute significantly due to the quadratic weighting of their distance. Thus, IM serves as a robust indicator of the underlying activity or variability of the phenomenon under study. The normalization process minimizes the impact of heterogeneity in the analyzed patterns and ensures that the sum of the occurrence values in each line of COM equals 126. This, in turn, ensures that IM is not biased by variations in the overall magnitude of the COM values, allowing for a more consistent and interpretable measure of dispersion26. The IM measurement serves as a valuable statistical tool for estimating the sample activity; increasing IM values correspond to increased sample activity. Absolute value of differences (AVD) The absolute or average value of differences (AVD) is the first moment analogous to IM. The AVD process quantifies a measurement relative to the absolute intensity leap value between successive patterns, specifically between consecutive patterns \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {I}_k$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\hbox {I}_{k+1}$$\end{document}. It operates on the principle that the accumulation of differences is the primary information sought, and is defined as27:3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \textrm{AVD} = \sum _{i}^{}\sum _{j}^{} \frac{\mathrm{{COM}}(i,j)}{\sum _{m}\mathrm{{COM}}(i,m)}|i-j|. \end{aligned}$$\end{document}Taking the absolute highlights the principle that the AVD method is fundamentally concerned with the magnitude of intensity changes between successive patterns. For a uniform intensity distribution, the AVD and IM yield similar results. However, a key distinction lies in their sensitivity to changes: while AVD uses absolute differences, IM employs squared differences. This means that the IM amplifies the contribution of larger deviations in the THSP values making it more sensitive to more significant variations. By examining AVD and IM together, a more comprehensive understanding of the sample’s activity can be achieved; AVD provides a balanced measure of overall variability, while the IM emphasizes larger and more pronounced changes. This dual analysis offers deeper insights into the underlying dynamics of the phenomenon under study8,27. Motion history image (MHI) Motion history image (MHI) is a visual representation commonly used in computer vision and image processing to capture and display the temporal evolution of motion in a sequence of images. The concept of MHI is a general graphical and static image that encapsulates the motion history of objects, portraying their paths as they evolve. This idea can be expanded to speckle patterns, providing a useful indicator of the sample activity8,28,29. To obtain MHI, the process begins with deriving the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_l(i,j)$$\end{document} matrix for each of the N speckle patterns. This matrix is obtained by subtracting a speckle pattern from the previous one, i.e., \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${S_{l}(i,j)=I_{l}(i,j)-I_{l-1}(i,j)}$$\end{document}, where, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$I_l(i,j)$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$I_ {l-1}(i,j)$$\end{document} represent the speckle patterns at moments l and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$l-1$$\end{document}, respectively. The resulting matrices are converted into binary types by applying a suitable threshold, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_\mathrm{{Threshold}}$$\end{document}, achieved upon considering the inherent reflectance of the sample and exposure adjustment:4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} {T_{l}(i, j)= {\left\{ \begin{array}{ll}1 & \text{ if } \left| S_{l}(i, j)\right|>S_\mathrm{{Threshold}}, \\ 0 & \text{ if } \left| S_{l}(i, j)\right| \le S_\mathrm{{Threshold}},\end{array}\right. }} \end{aligned}$$\end{document}Eventually, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\textrm{MHI}_l(i,j)$$\end{document} at instant l is defined as:5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} {\mathrm{{MHI}_{l}}=255 \sum _{k=0}^{N-1} T_{l-k} h_{k} \hspace{1mm},} \end{aligned}$$\end{document}where, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${h_{k}=\frac{N-k}{M}}$$\end{document} is the weighting factor that resembles the age of the pattern and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M = N(N +1)/2$$\end{document}. According to Eq. 5, the resulting MHI is a static pattern where each pixel’s intensity represents the cumulative motion history at that location. Due to incorporation of the age-weighting factor, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$h_k$$\end{document}, higher intensity values indicate areas that have experienced recent motion, while lower intensity values correspond to regions with less or no recent motion. Therefore, MHI depends on the number of patterns, the threshold, and the weighting of the subtractions. Toward extracting the utmost and proper information, these factors can be adjusted relevant to the speed of the phenomenon to be studied. In this definition, it is assumed that the speckle patterns are acquired with an 8-bit digital camera. Temporal intensity auto-correlation function The second-order intensity autocorrelation function, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$g_2(\tau )$$\end{document}, is a fundamental statistical measure used to quantify temporal fluctuations in the speckle patterns. It provides insight into the degree of correlation between intensity values measured at two times. It is widely used in the LSA, where direct tracking of individual scatterers is not feasible, which is exactly the condition of the present sample. The general form of the autocorrelation function is defined as30:6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} g_2(\Delta t) = \frac{\langle I(t) ~ I(t + \Delta t) \rangle }{\langle I(t) \rangle ^2}, \end{aligned}$$\end{document}where, I(t) refers to the measured speckle intensity at time t, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$I(t + \Delta t)$$\end{document} denotes the intensity recorded at a later time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t + \Delta t$$\end{document}. The variable \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta t$$\end{document} represents the delay time or lag time, which determines the temporal scale over which the similarity between intensity values is evaluated. The notation \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\langle ~ \cdot ~ \rangle$$\end{document} represents an averaging operation, which may be performed in the temporal domain by averaging over multiple time points at a single pixel location, or in the spatial domain by averaging the intensity values across a group of pixels within a single image frame. The function \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$g_2(\Delta t)$$\end{document} describes how similar the speckle intensity remains after a time shift of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta t$$\end{document}. At \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta t = 0$$\end{document}, where no delay is introduced, the correlation is maximal and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$g_2(0)$$\end{document} reaches its peak value. As \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta t$$\end{document} increases, the speckle pattern gradually decorrelates due to scatterer motions, causing \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$g_2(\Delta t)$$\end{document} to decay. In static or slowly evolving samples, the decay is minimal or slow, whereas in dynamically active samples, such as suspensions, gels, or powders undergoing microscopic rearrangements, the decay is more rapid. The shape and rate of decay of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$g_2(\Delta t)$$\end{document} provide valuable information about the internal dynamics of the sample, such as the characteristic decorrelation time, which can be further analyzed to infer mechanical or transport properties. Drying rate and surface roughness measurement In order to validate the LSA results, we also perform measurements on the drying rate and surface roughness of the powder samples. To obtain a quantitative measure of moisture loss, gravimetric monitoring is performed. Cornstarch powder samples are placed on a high-precision balance (Sartorius, BL 120S) with an accuracy of 10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^{-4}$$\end{document} g, and their mass is recorded throughout the drying process. Since the reduction in mass predominantly results from water evaporation, this method provides a direct and reliable measure of moisture loss over time. The sample mass M(t) is monitored as a function of time during the decrease in the relative humidity of the powder from 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} under stable ambient temperature conditions. The drying rate is calculated from the mass time series as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{\Delta M(t)}{\Delta t}$$\end{document}. These measurements serve as the reference baseline for LSA results. To characterize changes in the surface geometry of the samples during the drying process, a 3D optical profilometer (KEYENCE VR-6000) is employed. This profilometer captures the height distribution with approximately 0.1 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu$$\end{document}m precision of the sample surface, providing maps of topographical evolution over time. Given that laser speckle behavior is sensitive to surface roughness and geometry, profilometry is used to monitor the effect of moisture loss on the physical structure of the powder bed. These surface measurements enable the identification of microscale morphological changes, including the emergence of localized surface fractures. They serve as a complementary modality to the LSA method, allowing correlation between optical activity and physical deformation at the powder surface. Results and discussion The LSA analysis results are presented in Figs. 2, 3, 4, 5 and 6. Figure 2 shows the THSP matrices corresponding to the powder sample during a humidity transition from 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}. The stacks of speckle patterns are acquired at six different times; 1, 10, 20, 30, 90, and 100 minutes after the onset of the drying process. The corresponding calculated THSP matrices are shown in Figs. 2(a-f). THSP matrices are created by considering 500 randomly chosen pixels throughout a speckle pattern as a column, and placing the similar columns of a collection of 200 speckle patterns next to each other, resulting in THSP matrices of size \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$500 \times 200$$\end{document}. Sharp fluctuations in the intensity along each row reflect temporal activity at a specific spatial location. When such fluctuations appear consistently across multiple rows, they indicate widespread activity within the sample, which is caused by internal changes such as moisture loss. These variations manifest as horizontal bright and dark streaks in the THSP matrices. With the progression of the drying process and the gradual evaporation of moisture, temporal intensity variations in the THSP matrices reduce, indicating a reduction in internal dynamic activity within the powder structure. This reduction in activity corresponds to the decreasing water content in the sample, leading to a more stable scattering environment. In cases of very high level activity, the THSP pattern turns into, even, an ordinary speckle pattern, in which the bright lines cannot be followed and the structure looks like a random light field. However, this is not the case in most of our results. Fig. 2THSP matrices illustrating the powder dehumidification process during a humidity transition from 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} at different time points: (a) 1 min, (b) 10 min, (c) 20 min, (d) 30 min, (e) 90 min, and (f) 100 min, after the start of the experiment. THSPs are created by considering 500 randomly chosen pixels throughout a speckle pattern as a column, and placing the similar columns of a collection of 200 speckle patterns next to each other. Fig. 33D plots and 2D maps of COM matrices obtained during a humidity transition from 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} at different time points: (a) 1 min, (b) 10 min, (c) 20 min, (d) 30 min, (e) 90 min, and (f) 100 min, after the beginning of drying. Reference level and comparison level correspond to the gray-level intensities i and j as defined in Eq. 1, respectively. THSP matrices illustrating the powder dehumidification process during a humidity transition from 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} at different time points: (a) 1 min, (b) 10 min, (c) 20 min, (d) 30 min, (e) 90 min, and (f) 100 min, after the start of the experiment. THSPs are created by considering 500 randomly chosen pixels throughout a speckle pattern as a column, and placing the similar columns of a collection of 200 speckle patterns next to each other. 3D plots and 2D maps of COM matrices obtained during a humidity transition from 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} at different time points: (a) 1 min, (b) 10 min, (c) 20 min, (d) 30 min, (e) 90 min, and (f) 100 min, after the beginning of drying. Reference level and comparison level correspond to the gray-level intensities i and j as defined in Eq. 1, respectively. To better understand the changes in the sample activity during the dehumidification process, we calculate the associated COM for each case. Figs. 3(a-f) show the COM matrices in 3D and 2D modes during a humidity transition from 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} at different times 1, 10, 20, 30, 90, 100 min, respectively. The COM results represent these features: the decrease of the powder dehumidification process leads to further COM values becoming more concentrated around the main diagonal, and a more increased peak of the data distribution. Therefore, it seems COM is a suitable indicator for monitoring moisture, as, according to the explanations presented in the previous section, the more spread of data further away from the main diagonal, the higher the sample activity.Fig. 4The MHI images generated from a collection of 200 successive speckle patterns, during a humidity transition from 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} at different time points: (a) 1 min (start of the experiment), (b) 10 min, (c) 20 min, (d) 30 min, (e) 90 min, and (f) 100 minutes after the beginning of drying. The MHI images generated from a collection of 200 successive speckle patterns, during a humidity transition from 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} at different time points: (a) 1 min (start of the experiment), (b) 10 min, (c) 20 min, (d) 30 min, (e) 90 min, and (f) 100 minutes after the beginning of drying. Figure 4(a-f) show the MHIs generated from the same stack of the speckle patterns used for the THSP analysis and the other parameters derived from THSP matrices during a humidity transition from 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} at different times 1, 10, 20, 30, 90, 100 min, respectively. As seen in the figure, it is evident that during the drying process, the values of the MHI elements decrease and areas of blue color become more pronounced. This, according to the definition of MHI and the color bar of Fig. 4, indicates a decrease in the activity of the sample. Also, the results show that after a certain time, the MHI matrix remains almost unvaried, indicating that the sample has reached a static state, i.e., the powder is dried.Fig. 5The variations of (a) IM and (b) AVD, as speckle analysis quantitative indicators during the powder dehumidification process under three relative-humidity ranges (90-45\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}, 70-50\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}, and 50-45\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}) at different time points. The variations of (a) IM and (b) AVD, as speckle analysis quantitative indicators during the powder dehumidification process under three relative-humidity ranges (90-45\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}, 70-50\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}, and 50-45\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}) at different time points. To represent the changes in a more quantitative fashion, IM and AVD for three relative-humidity ranges (90-45\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}, 70-50\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}, and 50-45\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}) are also computed and shown in Figs. 5(a) and 5(b), respectively. Both parameters exhibit similar trends over time, consistent with their mathematical definition and their shared dependence on the underlying activity of the system. Under all conditions, both metrics exhibit a decreasing trend over time, reflecting a continuous reduction in sample activity driven by progressive moisture evaporation. Among the tested humidity ranges, the 90-45\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} condition, representing the highest initial humidity, affords the largest initial IM and AVD values, indicating elevated activity at the beginning of the process. However, due to the higher moisture content, the drying occurs more slowly, and the activity persists for a longer period before stabilizing. In comparison, for the 70-45\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} and 50-45\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} ranges, the activity drops more rapidly and reaches a steady state within the first few minutes, suggesting faster drying dynamics under lower initial humidity levels. These observations highlight the strong dependence of drying rate and activity evolution on the initial moisture content. The similarity in their behavior stems from the fact that IM and AVD are closely related statistical measures. While the IM emphasizes larger deviations due to its quadratic weighting of distances (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| i - j \right| ^2$$\end{document}), the AVD provides a more balanced measure of variability by using absolute differences (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left| i - j \right|$$\end{document}). This difference in sensitivity means that the IM is more responsive to significant variations, whereas the AVD offers a smoother representation of overall changes. By analyzing the parameters together, a more nuanced understanding of the dehumidification process is achieved. The steady-state values correspond to a low-activity regime with strongly reduced speckle fluctuations. Although the final humidity levels are not identical, this behavior reflects suppressed moisture-driven dynamics rather than equal moisture content. All acquisition parameters are kept identical, ensuring that the observed differences arise from intrinsic sample dynamics, with minor variability under controlled conditions.Fig. 6(a) Temporal intensity autocorrelation function (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$g_{2}(\Delta t)-1$$\end{document}) as a function of the lag time (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta t$$\end{document}) for four drying durations of cornstarch powder exposed to a humidity transition from 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}. Dashed curves represent exponential fits of the form \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\exp (-\Delta t/\tau ) + C$$\end{document}, and the fitted decorrelation times (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tau$$\end{document}) are shown next to their corresponding curves. (b) Temporal intensity autocorrelation function as a function of drying time for a fixed lag time of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta t = 0.5\textrm{s}$$\end{document}. The curves correspond to four relative humidity ranges: 29\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} (reference), 55-29\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}, 75-35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}, and 90-35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}.Fig. 7(a) Mass over time and (b) drying rate versus time for a cornstarch powder sample, incubated at 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} relative humidity, exposed to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} relative humidity. (a) Temporal intensity autocorrelation function (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$g_{2}(\Delta t)-1$$\end{document}) as a function of the lag time (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta t$$\end{document}) for four drying durations of cornstarch powder exposed to a humidity transition from 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}. Dashed curves represent exponential fits of the form \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\exp (-\Delta t/\tau ) + C$$\end{document}, and the fitted decorrelation times (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tau$$\end{document}) are shown next to their corresponding curves. (b) Temporal intensity autocorrelation function as a function of drying time for a fixed lag time of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta t = 0.5\textrm{s}$$\end{document}. The curves correspond to four relative humidity ranges: 29\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} (reference), 55-29\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}, 75-35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}, and 90-35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}. (a) Mass over time and (b) drying rate versus time for a cornstarch powder sample, incubated at 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} relative humidity, exposed to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} relative humidity. During the drying process, we evaluate the temporal intensity correlation function between frames separated by a lag time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta t$$\end{document}. In this research, we analyze the quantity \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$g_{2}(\Delta t) - 1$$\end{document}, which emphasizes the fluctuating component of the speckle intensity. Subtracting 1 removes the static offset associated with speckle contrast and facilitates direct comparison of the correlation curves across different drying stages30,31. Figure 6(a) shows the plots of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$g_{2}(\Delta t)-1$$\end{document} as a function of the lag time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta t$$\end{document} for four different drying durations of cornstarch powder. In this experiment, the sample is first equilibrated at 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} relative humidity and then exposed to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} relative humidity. For each drying stage, successive speckle patterns are recorded and their corresponding \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$g_{2}(\Delta t)-1$$\end{document} curve is computed. At early drying times, for example 2 min after the data acquisition (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\diamondsuit$$\end{document}) the curves decay rapidly with increasing \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta t$$\end{document}, indicating that the speckle pattern changes quickly from frame to frame. In contrast, after extended drying, the correlation curves remain close to their initial values, reflecting much slower temporal variations in the speckle intensity. To quantify these differences, each correlation curve is fitted with an exponential model of the form \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\exp (-\Delta t/\tau ) + C$$\end{document}, where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tau$$\end{document} is a characteristic decorrelation time, at which \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$g_{2}$$\end{document} reaches \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{1}{e}$$\end{document} of its initial value. The lower values of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tau$$\end{document} correspond to faster decorrelation and therefore stronger speckle activity, while the higher values indicate slower decorrelation and a more stable speckle pattern. The characteristic decorrelation times, displayed next to their corresponding curves in Fig.6(a), experience a clear increase as drying progresses. A complementary view of the drying dynamics is obtained by analyzing the temporal intensity autocorrelation at a fixed lag time. Figure 6(b) shows the temporal intensity autocorrelation function \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$g_{2}(\Delta t)-1$$\end{document} evaluated at the lag time of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta t = 0.5\,\textrm{s}$$\end{document} as a function of drying time for four relative humidity ranges: 29\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} (reference), 55-29\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}, 75-35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}, and 90-35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document}. This representation highlights how fast the speckle pattern evolves at this short time scale during different stages of drying. At early drying times, the values of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$g_{2}(0.5 \textrm{s}) - 1$$\end{document} are relatively low, indicating rapid temporal fluctuations in the intensity of the speckle pattern. The magnitude of these fluctuations also depends on the initial humidity difference; larger relative humidity gradients lead to stronger variations. As drying progresses, the values of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$g_{2}(0.5 \textrm{s}) - 1$$\end{document} become progressively higher, reflecting a gradual reduction in the rate of intensity changes in the speckle pattern. Samples subjected to larger humidity gradients, e.g., 90-35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} relative humidity, exhibit a slower overall evolution of the correlation coefficient, indicating that pronounced temporal speckle fluctuations persist over a longer part of the drying process. In contrast, samples with smaller humidity differences reach higher correlation values at earlier drying times, consistent with faster stabilization of the speckle pattern. To validate the remote optical measurements and assess the accuracy of the LSA, independent gravimetric drying experiments are performed. Figure 7(a) shows the sample mass M(t) for cornstarch powder dried from 90\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} to 35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\%$$\end{document} relative humidity under stable ambient conditions. The mass decreases rapidly at early times, consistent with the evaporation of surface water. As drying proceeds, the available moisture within the powder reduces, resulting in progressively slower mass loss and a smooth approach to a near-equilibrium value after approximately 1000 min. Figure 7(b) displays the drying rate, calculated from the time series of the mass as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta M / \Delta t$$\end{document}. The drying rate is highest at the start and decreases sharply within the first \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sim$$\end{document}100 min. After this stage, the drying rate gradually decreases and eventually reaches a low, nearly stable level. This trend reflects a characteristic two-stage drying behavior: an initial rapid moisture-loss phase driven by surface evaporation, followed by an extended slow phase governed by diffusion-limited transport of water from the interior of the powder.Fig. 8Profilometry images showing occurrence of a micro-fracture during the drying of cornstarch powder at 1 min, 10 min, 20 min, 50 min, and 240 min. Profilometry images showing occurrence of a micro-fracture during the drying of cornstarch powder at 1 min, 10 min, 20 min, 50 min, and 240 min. As an additional verification of the optical measurements, the evolution of the powder surface is examined using 3D optical profilometry. This technique directly captures changes in surface height and morphology, providing an independent view of how moisture loss affects the physical structure of the powder bed. Figure 8 shows profilometry height maps of cornstarch at drying times of 1, 10, 20, 50, and 240 min. During the early stage (1–20 min), the surface exhibits only mild topographical variations. By 50 min, different micro-fractures and localized surface discontinuities become clearly visible, reflecting mechanical contraction and rearrangement of the powder as water evaporates. At longer drying times, e.g., 240 min, the overall surface morphology changes only modestly, indicating that the major deformation associated with drying occurs predominantly within the first hour. The gravimetric and profilometry results provide independent confirmation of the trends revealed by LSA. All methods consistently show a rapid initial stage followed by a slower, stabilized phase. From a physical perspective, this two-stage behavior can be attributed to a transition from a surface-controlled evaporation regime to a diffusion-limited moisture transport regime within the powder, which is qualitatively consistent with Fick-type diffusion32. Within this framework, the evolution of the dynamic speckle patterns can be interpreted as an indirect signature of the underlying transport processes, where higher speckle activity corresponds to the rapid mass-loss stage, and its subsequent decline reflects the progressive suppression of moisture-driven internal rearrangements during drying. This agreement highlights the reliability of our optical approach and underlines its potential as a simple and powerful tool for monitoring drying dynamics in powders. It is remarkable that although cornstarch powder is used in this study as a model sample, the applicability of the method is not limited to this material and can be extended to a wide range of powders that undergo moisture-driven or microstructural changes. The performance of the method depends on key factors such as the optical properties of the sample, e.g., scattering and absorption, particle size and packing structure, and the level of internal dynamics during the process. Materials with sufficient scattering and active internal rearrangements are expected to produce higher-contrast and more informative speckle patterns, whereas highly absorbing or weakly dynamic systems may lead to reduced sensitivity. Although the present experiments are performed under controlled laboratory conditions, the proposed method is based on a relatively simple and compact optical setup, which makes it potentially suitable for practical implementation. In real processing environments, additional factors such as mechanical vibrations, ambient light fluctuations, and environmental variations may influence the LSA. Nevertheless, such limitations can be mitigated through appropriate system design. These include mechanical stabilization, controlled illumination, and signal normalization strategies. Despite the aforementioned limitations, many industrially relevant powders, including food and pharmaceutical materials, share similar characteristics to cornstarch in terms of scattering behavior and moisture sensitivity. Therefore, the method can be considered broadly applicable, although calibration or adjustment may be required for specific systems. Within these considerations, LSA can be valuable in processes involving drying, conditioning, or rehydration of food, pharmaceutical, and industrial powders, where real-time and non-contact assessment of their internal activity is essential for ensuring product quality and process control. Conclusion This study shows that laser speckle analysis (LSA) is an effective and non-contact method for monitoring the drying process in hygroscopic powders. By analyzing speckle patterns generated at the powder surface, we showed that temporal speckle fluctuations provide a sensitive optical signature of the internal activity of the material during drying. Several statistical analysis methods are evaluated, including the time history of speckle patterns, the co-occurrence matrix, the motion history image, the inertia moment, the absolute value of differences, and the temporal intensity autocorrelation function. All of these quantities exhibited a consistent behavior: high speckle activity at the beginning of drying, followed by a progressive reduction as moisture is removed. The autocorrelation analysis further revealed an increase in decorrelation time with drying duration and a clear dependence on the imposed humidity gradients, indicating that LSA can discriminate between different drying conditions and stages. The speckle-derived trends are further validated through independent gravimetric measurements and 3D optical profilometry. Both methods exhibited drying behaviors consistent with the optical observations, confirming that LSA provides an accurate representation of the material’s dynamical and structural evolution during moisture loss. These results highlight LSA as a practical and accessible technique for characterizing drying in powders. Its non-contact operation, simple setup, and quantitative output make it well-suited for use beyond laboratory conditions. The method can be readily adapted for real-time monitoring and control in several processes involving drying, conditioning, or rehydration of food, pharmaceutical, and industrial powders.