Application of Image Processing for Inspection of Pill Production Process

摘要

Inspection of the external appearance of the pill is one of the methods to check the basic quality according to the standard of Food and Drug Administration (FDA) such as chipped pills, inappropriate sizes, and unclear letter printed on pills. This paper proposes an application of image processing for inspection of pill production process. In this study, the Image Processing principal including the Grayscale Method, Threshold Method, OTSU Method, Bounding Box Method and Geometric Algorithm was adopted to process images from normal webcam camera. In addition, the Microsoft Visual C# 2008 was used for processing the signals by analyzing the output signals of webcam camera. In the experiment, the perfect pill was determined by a cross-sectional area not less than 98% of the standard pill in units of milligrams (mg) in FDA standard. The experiment was divided into three cases: inspection 100 perfect pills, 100 imperfect pills and 100 pills in 1:1 ratio between the perfect pills and the imperfect pills with the speed for inspecting and distinguishing of 70 pills per minute. From the experimental result of three cases, the error rates were 0.6%, 1.2% and 1.0%, respectively, while the average of process error was 0.93%.