Research On Stamping Parts Inspection Based On Machine Vision

Manufacturing is the pillar industry in China. With the global manufacturing center transferring to our country, higher quality, higher producing efficiency, lower cost for manufacturing and so on are proposed by market. Hence domestic manufacturing industry is in an urgent need for improving manufacturing and detecting techniques. But many domestic manufacturing enterprises still adopt manual inspection, which cost biggish manpower and administration expense, and cannot make a better balance between inspecting quality and efficiency. This problem has greatly restricted the development of our manufacturing industry. Machine vision based inspection is a new intelligent inspecting technology developed in recent years. It use industry equipments substituting human eyes to conduct products inspection. The excellent features like:objectivity, efficient, accurate, non-touching, low cost, easy maintenance and so on, make it become an ideal technical method in production monitoring stage currently.The research contents in this paper is the computer stamping inspection based on machine vision. This study origins from a development project from a computer components supplier. The detecting target is the defects on the computer components, including shape defects: burr, dent; surface defects:scratch, dirt, white spot, oxidation and so on. Owning to the defects are abundant in variety and the size is so small that manual detection cannot ensure a higher inspecting quality and efficiency because of the inherent human limits. The enterprise wants to develop an efficient online inspecting system to replace manual detection and thanks to excellent features mentioned above, the machine vision based inspecting technique become an ideal method to meet the require proposed by the enterprise.Based on the analysis of detection requirements proposed by the enterprise, a serious study has conducted and has accomplished following works:1. Aiming at the defects to be inspected, the optical part of the inspecting system has been design. Use LED as light source, a abundant of experiments have been conducts to determine the lighting ways of different defects. This article also illustrates relations between detecting accuracy and camera resolution and make sure the resolution of the camera;2. Based on the analysis of different defects image features, a low pass image differencing the original method is proposed to segment the defects. Experiments show that, this method can achieve an ideal defects segmentation results and better than conventional image segmentation methods.3. Analyse defects region features, introduce conventional region feature judging methods, and finally determine adopt inner-circle feature to distinguish different type of defects;4. Based on the algorithm this article proposed, a reasonable detecting software has been developed. This software system is developed by VS2008, some detection algorithms used the optimization algorithm provide by MVT Halcon, realize image data acquisition, save and quantificationally analyze defects.The novelty of the article is based on the type of defects and the demands of precision, proposed an vision-based defects inspecting system design program, combining Mathematical morphology and blob analysis methods to achieve inspecting purpose. The results shows that the practicality of this detecting system and the processing speed can fulfill on line detection.