Fast Sphericity and Roundness approximation in 2D and 3D using Local Thickness
Pawel Tomasz Pieta, Peter Winkel Rasmussen, Anders Bjorholm Dahl, Anders Nymark Christensen · 2025
Read Fast Sphericity and Roundness approximation in 2D and 3D using Local Thickness by Pawel Tomasz Pieta, Peter Winkel Rasmussen, Anders Bjorholm Dahl, Anders Nymark Christensen (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-05808.txt.
Introduction
Pawel Tomasz Pieta and a team of co-authors speed up measures of how round and spherical objects look in images, in two thousand twenty-five. Sphericity and roundness are basic shape scores for cells and particles in microscopy. Their strict definitions are costly to compute, and modern image sets keep growing in size.
Their method reuses the output of a local thickness algorithm, which gives the width of an object at each point. For sphericity, objects are modeled as ellipses or spheroids sized by their mean thickness. For roundness, thickness values along the edge stand in for a costly curvature measure. The shortcuts stay close to the exact scores while running far faster.
Catalog record
| CREATOR | Pawel Tomasz Pieta, Peter Winkel Rasmussen, Anders Bjorholm Dahl, Anders Nymark Christensen |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 32 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-05808
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Fast Sphericity and Roundness approximation in 2D and 3D using Local Thickness Pawel Tomasz Pieta, Peter Winkel Rasmussen, Anders Bjorholm Dahl, Anders Nymark Christensen Abstract Sphericity and roundness are fundamental measures used for assessing object uniformity in 2D and 3D images. However, using their strict definition makes computation costly. As both 2D and 3D microscopy imaging datasets grow larger, there is an increased demand for efficient algorithms that can quantify multiple objects in large volumes. We propose a novel approach for extracting sphericity and roundness based on the output of a local thickness algorithm. For sphericity, we simplify the surface area computation by modeling objects as spheroids/ellipses of varying lengths and widths of mean local thickness. For roundness, we avoid a complex corner curvature determination process by approximating it with local thickness values on the contour/surface of the object. The resulting methods provide an accurate representation of the exact measures while being significantly faster than their existing implementations. 1 Introduction The analysis of segmentation masks is a standard part of image processing pipelines. Especially within microscopy, the morphology of segmented objects is a frequent research target, requiring a wide range of descriptors [ 61 , 8 , 15 , 9 , 25 , 30 , 2 , 34 , 29 ] . The simplest statistics and shape measures can be seamlessly extracted using standard image-processing libraries [ 10 , 51 , 54 ] . While these are useful and fast to extract, they are often not descriptive enough for a detailed analysis of the shape of the object. One of the more advanced shape measures is local thickness [ 26 , 17 ] , defined as the radius of the largest circle/sphere that fits in the object at any point inside it. Although its calculation is relatively complex, recent advancements have shown that it can be efficiently accelerated while maintaining high accuracy [ 14 ] . Figure 1 : Visual chart for estimating sphericity and roundness from 2D silhouettes, adapted from [ 32 ] . Similar to local thickness — sphericity and roundness can also serve as object descriptors, focusing much more on its shape than its dimensions. Originating from geology [ 56 ] , they are widely used in the study of rock/granular particle shape [ 3 , 6 , 21 , 46 , 50 , 52 ] . However, their descriptive power extends their applicability beyond geology, finding use both within bio-imaging [ 1 , 22 ] and other image-based studies [ 23 , 36 , 43 , 55 , 27 ] . Both sphericity and roundness capture how close an object is to a perfect circle/sphere, but they describe two different, complementary characteristics. Sphericity depends primarily on the elongation of the object, while roundness captures the roughness or inconsistency of its edges. As a result, an elongated but smooth object will have low sphericity and high roundness, while a compact and rugged object will have high sphericity but low roundness ( Fig. 1 ). Many implementations of the two measures have been suggested, both for 2D and 3D data [ 20 , 59 , 38 , 63 , 64 , 11 , 62 , 13 ] . Out of the two, sphericity is easier to implement, typically requiring an extraction of an object’s perimeter in 2D and mesh in 3D. Calculating roundness continues to be a challenge, especially in 3D, as in the most strict form it requires collecting radii of curvature of all object corners and ridges. While some approximate methods for calculating roundness also exist [ 18 , 48 ] , their implementations are rarely available. Overall, even the most popular implementations of the two measures are relatively complex and unfit for modern data-heavy analysis. They tend to perform adequately when evaluating a handful of objects, but often struggle to accommodate large groups of objects at once. We propose a fast and scalable method for extracting the sphericity and roundness of objects from 2D and 3D masks based on their local thickness values. Noticing the parallels between the sphere-based definition of local thickness and inputs needed for calculating the two measures, we simplify their calculation to a set of fast operations that can easily be applied to all image objects at once. For sphericity, we model the objects as ellipses (in 2D) and spheroids (in 3D) with a width of mean local thickness and volume-dependent height. This approach lets us avoid the meshing step with a minimal loss in accuracy for the majority of the most common object shapes. For roundness, we extract local thickness values from a simplified perimeter/shell of the object and treat them as the curvature radii. While not exact, the resulting measure exhibits high levels of correlation to the original measure. Our implementation is directly available for use in Python, as a pip-installable package under https://github.com/PaPieta/fast_rs . 2 Background The most established definitions of sphericity and roundness were formulated by Wadell [56] , Wadell [57] , Wadell [58] . Both measures were designed to describe the geometry of 3D objects, such as rocks, pebbles, or particles. While sphericity was a 3D definition from the start, roundness was only defined for 2D object projections. Initially, they were assessed primarily through physical measurements, but the complexity of these assessments led to the creation of charts that allowed for approximate visual estimation [ 31 , 32 , 45 ] ( Fig. 1 ). Later developments brought more sophisticated measurement techniques, such as using specialized tools [ 5 , 24 , 7 ] or performing image-based processing [ 60 , 16 ] . The simplicity and potential of the image-based approach, as well as advancements in 3D imaging, have fueled the research and development of new roundness and sphericity algorithms (detailed in Sec. 3 ). These methods often used the sphericity and roundness charts as baselines for evaluating and benchmarking their performance. 2.1 Sphericity background Sphericity is defined in 3D as the ratio of the surface area of a sphere, of the same volume as the particle, to the actual surface area of the particle [ 56 ] : \mathcal{S}_{3D}=\frac{\pi^{\frac{1}{3}}(6V)^{\frac{2}{3}}}{S} (1) where V is the volume of the particle/object and S is its surface area. A direct translation of 3D sphericity to 2D suggests the use of perimeter sphericity [ 12 , 33 , 4 ] , or ISO(2008) circularity [ 28 ] : \mathcal{S}_{2D,P}=\frac{P_{c}}{P_{o}} (2) where P_{c} is the perimeter of a circle having the same projected area as the object, and P_{o} is the perimeter of the object. However, it is not clear which of the many proposed specificity definitions [ 42 , 47 ] was used for preparing the baseline visual approximation charts. A recent investigation [ 63 ] suggests that Krumbein and Sloss [32] used an alternative definition of width-to-lenght ratio sphericity in his charts ( Fig. 1 ): \mathcal{S}_{2D,WL}=\frac{d_{2}}{d_{1}} (3) where d_{1} and d_{2} are the length and width of the object. 2.2 Roundness background Roundness is defined in 2D as the ratio of the average curvature radius of all corners of a particle to the radius of the maximum inscribed circle [ 56 ] : \mathcal{R}_{2D}=\frac{\sum^{N}_{i=1}r_{i}}{NR} (4) where r_{i} is the curvature radius of the i -th corner, R