Color Defect Detection of Pharmaceutical Capsules using Histogram based Thresholding

Abstract

In this paper, a new method of inspecting pharmaceutical capsules aesthetic defects is proposed using color image processing. After an image is taken, median filtering is performed. Then the region based statistics is calculated to acquire the largest area of pharmaceutical capsule from the image. After that RGB image is transferred into R plane, G plane and B plane. Then histogram of each plane is calculated. Further by setting thresholding value of each plane, mask of each plane is determined. Color segmentation of image is achieved by combining mask of each plane. Experiments are conducted with the quality inspection of pharmaceutical capsules. Time is also taken as major parameter of the system. Experiment results show the reliability and feasibility of proposed method as time and quality control are two main parameters in industry. The proposed paper present an application to detect defect related to color of pharmaceutical capsules which can be adapted to other practical applications for automated visual inspection using image processing techniques.