Development Of Bottle Cap Defect Detection System Based On Machine Vision

As the food safety problems frequently happen, the consumer has higher requirement about the quality of the food packaging,so the plastic anti-theft caps at the factory, usually through rigorous quality testing.such as the print pattern is clear correctly, presence of stains, roundness deformation, lack of material, material overflow, presence of anti-theft ring.In order to avoid shortcomings that are caused by production line artificial recognition that it easily fatigue and easily make mistakes, it need to develop a set of detection system that applied to production line for automatic idenfication on the bottle caps.Based on practical engineering requirements, this article mainly had completed the study of algorithm design as follows:1. Pattern printing defect detection. Detection of beverage closures whether there is the print pattern, the color of the printed pattern is correct, and if there is a obvious defect pattern. However, in real-time systems, the specific location of the defect does not need to be marked, so the paper based on statistical image features to be judged.2. Regional marking. The image of bottle caps mouth is divided into three zones: the mouth, the mouth two and cap side wall. Firstly, the use of the outer contour obtain contour bottle and mouth two contour extraction method based on Freeman chain code, and then through the contours of the image scaling principle to obtain the mouth, finally filling algorithm based on Freemanl chain code to fill bottle cap area.3. Inner surface stains detection.To identify the stains area: a mouth, the mouth two, bottle cap wall. Because of the three regions exist different brightness characteristics, so in this paper, the three regions used different threshold for sliding window to extract stain.However the existence of outline in outline of the edge area, so we should avoid the effects of edge profile to stain the recognition in the process of stain detection. In this paper, these areas use Sobel filter for edge detection.4. Bottle cap shape detection. Testing purpose is to find roundness deformation, lack of material or overflow, and other defects.This paper propose two methods of detection algorithm, you can choose the appropriate algorithm according to different situations in practical application.After finishing design and implement the system, the test results of the detection system show that the defect detection system can meet the basic requirements of the production line, and has been recognized by the user.