Resersh On Rapid Detection Algorithm Based On Machine Vision For Appearance Deficiency Of Coating Products

The appearance quality of coating product has a great influence on the protective performance and the decorative performance. Therefore, on-line inspection for the appearance quality of coating products is a very important link to ensure the quality of the products. The quality and efficiency of the traditional artificial detection method can not meet the production requirements. According to the requirements of the enterprise, our research group has developed an on-line detection system for the surface quality of coating products. Based on machine vision technology, this paper makes a deep research on the rapid detection algorithm for the appearance defects of coating products.In the detection of surface defects, after the analysis of problems of traditional detecting algorithm applied in coating products, a defect detection algorithm based on well- filter and self-subtraction of image is proposed, and it can reinforce character of defect, and improve the detection rate. In the stage of defect judgment, the method of defect screening based on open operation and the method of correcting distortion of the defect size caused by image acquisition are proposed.Image rapid matching algorithm is a basic algorithm of image registration and fine-grained difference analysis of image in the detection of deformation defects, and the matching performance and efficiency of the algorithm can determine directly the quality and efficiency of the deformation defect detection. Therefore, according to shape characteristics of the coating products, this paper proposes a shape-feature-based normalized cross correlation(SFB-NCC) matching algorithm. What’s more, the efficiency of the matching algorithm is significantly optimized by the shape feature insteading of the gray feature, image pyramid, window-matching condition and adaptive-step-size hop operation of column and row. Experimental results show that the algorithm not only has a high precision and good adaptability, but also has a high speed and can meet the needs of online detection.In the detection of deformation defects, based on SFB-NCC matching algorithm, image registration algorithm can correct distortion of the workpiece image, such as small rotation, translation, tension and so on. Fine-grained difference analysis of image is processed by slicing the large size image, matching sub images based on SFB-NCC matching algorithm, and obtaining the accurate difference image, so that the defects can be easily extracted.