The size and shape characterization of platelike particles using a dual projection imaging device is presented. Based on the published algorithm to estimate the three lengths of the particles from the image pairs (method 1), the average error on the shortest length is around 140% on a test set of particles with different sizes, shapes, and alignment. Therefore, two potential improvements are tested in a simulation setting. The first approach uses an additional projection coupled with an oriented bounding box to estimate the three lengths (method 2). The second approach uses a machine learning model to estimate the three lengths from two projections (method 3). It is shown that method 2 marginally increases the accuracy, while method 3 leads to a significant improvement with an average error of 33% for the shortest characteristic length of particles in the test set and even less for the other lengths. It is shown that this remaining error is caused by the particle alignment with respect to the cameras. Since platelets cannot always be automatically distinguished from quasi-equant particles due to alignment issues, the model was also trained to estimate the three characteristic lengths of quasi-equant particles. The training set and the test set, used for model training and validation, respectively, comprise particles that exhibit quasi-equant, needlelike, and platelike shapes. Several machine learning models are identified and optimized to predict the three particle lengths based on the two projections from the imaging device. Artificial neural networks were chosen because of their superior predictive performance.

materials (e.g., metal-organic frameworks) and catalysts, and the production of pharmaceutical, agrochemical, and food compounds. In all situations, particle sizing plays a key role in the design and control of particle properties and particle size and shape manipulation.Conventionally, a single characteristic length is used to describe particles, assuming a spherical geometry, or taking advantage of the concept of equivalent diameter. This assumption may be acceptable for irregular yet compact shapes, and is an established simplification functional to the description of various physical phenomena. [1] However, it is clearly incorrect when analyzing elongated or plate-like particles, i.e., with low sphericity, which are common crystalline products. [2] This leads to particle size and shape estimates that poorly match physical reality. [3][4][5] To describe both size and shape, at least two characteristic lengths are needed, which can be obtained through a number of 2D or 3D characterization techniques. [6][7][8][9] Such (offline and online) techniques, including micro-computed tomography (µCT), [10][11][12][13][14][15][16][17] holography, [18][19][20] structural light, [21] machine vision, [13,[22][23][24][25][26][27][28][29][30][31] confocal microscopy, [32] and surface imaging [33] have varying degrees of accuracy and complexity. Machine vision through image analysis (IA) has become an increasingly powerful tool, due to experimental simplicity, hardware improvements, and continual increases in available computational power. [8] For the online measurement of elongated, needle-like particles on the microscale, effective IA algorithms have been proposed and successfully used to track particle length and width distributions. This has enabled a better fundamental understanding of processes in particle technology. [34][35][36][37][38] By contrast, an experimentally validated methodology enabling the rapid and online measurement of cuboidal or plate-like particles has yet to be developed. Full 3D characterization is possible through µCT, but this comes with several downsides, such as extended scanning times and, in the case of powders, difficulties in particle segmentation. [14] 2D characterization by single projection imaging can be fast, but does not provide sufficient information to reliably extract characteristic lengths and shapes, whilst the accuracy of multiple projection imaging is compromised by particle orientation. [6,8] Other techniques, such as holography or surface imaging, are yet to be applied to, or are unsuited for online measurement purposes.Characterization of particle size and shape is central to the study of particulate matter in its broadest sense. Whilst 1D characterization defines the state of the art, the development of 2D and 3D characterization methods has attracted increasing attention, due to a common need to measure particle shape alongside size. Herein, ensembles of micrometer-sized cuboidal particles are studied, for which reliable sizing techniques are currently missing. Such particle...