Research on Defect Detection Technology of Tablets in Aluminum Plastic Package

ABSTRACT

Aluminum plastic foamed mask package is commonly used for the packaging of tablets in the pharmaceutical companies and defective tablets can be produced during the production of the packaging. This paper studies several key algorithms of defect detection for tablets in aluminum and plastic package and designs the auto- detection software which applies an image processing and recognition technology. Firstly, it uses a weighted filtering method of color images to pretreat the tablet images. Secondly, it classifies the tablets into color and white tablets, and adopts a significant target segmentation method based on the visual significance analysis in order to respectively segment two kinds of tablets from the background image. Then detects the edges of tablets by Canny operator. Finally, it adopts the connected component labeling algorithm to detect the tablets lacking of numbers and other defects including misshapen tablets and tablets with scratches by the SVM algorithm. Large amounts of experiment results demonstrate that the system has greater accuracy and higher stability and speed.

Keywords:

Color image filtering, connected component labeling, defect detecting, SVM algorithm, visual significance.