Capsule Deformation in Pharmaceutical Industry by a Non-Contact Metrology Algorithm

Metrology is the science and art of precision measurement, testing and evaluation. It has a vital role in technological development. In this paper, a Non-contact metrology based new technique is proposed to automatically distinguish the qualified and unqualified capsules manufactured in the pharmaceutical industry based on image processing technique. The existing algorithms for the same purpose are work on the principle of pixel to pixel comparison with a fixed window size of neighborhood. When the image is scaled or translated, the pixel to pixel comparison is difficult to achieve. A small effort is, therefore, made to solve this problem using a new technique for automatic window size determination. Gaussian filter is applied for smoothing and non-linear filter is applied for re-smoothing as a preprocessing step. In addition, edges are detected by canny edge detector and Gabor energy filters. Results obtained reveal that the proposed technique for automatic window size determination is a good and efficient solution with the use of Hausdroff distance, through which a binary comparison is performed after segmentation and thresholding.