Development of Image Analysis Calibration Procedure for in-Situ Crystal Size Measurement

Crystallization is a process commonly used in industries such as pharmaceuticals and agriculture, and the formation of crystals of a specific morphology, such as a certain length, are often crucial to the formation of products. Offline measurements via image analysis or laser diffraction are methods to accurately measure the crystal sizes. However, the infrequency of sampling and sample handling can disturb the dynamics of the crystallization system. A non-invasive solution to address data scarcity in particle size is to obtain crystal length by an in-situ camera known as a Particle Vision Monitor (PVM), followed by image analysis algorithms to obtain a length distribution. However, particles frequently overlap or are of different morphologies, leading to inaccuracies in the current algorithm. In order to improve the image analysis of crystal length for a specified crystal system, a recirculation loop was devised alongside a new algorithm utilizing MATLAB in conjunction with ImageJ. Once the images have been obtained, the process screens parameters such as the minimum area, the background subtraction, the minimum aspect ratio, and the circularity parameter to remove unnecessary noise. Once a new distribution of the particles is obtained from the chosen parameters, it is compared to an offline measured distribution as a way to calibrate the parameters to fit the offline measurement. An objective function, which quantifies the absolute error of the obtained distribution compared to the offline one, is calculated. New parameters values are then selected in order to minimize the error of the objective function. Currently, various relationships between concentration of the crystal system and each parameter are being derived in order to create a system that can output a more accurate length distribution for a selected concentration.