| Defect segmentation is the most important part of the rolled strip surface defect inspection. Whether the high efficiency and precision will work on the following classify and identification, so as to the whole system performance. As the cold rolled strip surface images have different defects, background is influenced by a lot of factors such as illumination uniformity and abnormal of the texture. It is difficult to find a suitability solution to cold rolled strip surface defect segmentation.This paper expatiates and implements a surface defect segmentation algorithm based on edge, and the experiment results are present. Then this paper introduces a defect segmentation algorithm based on artificial immune algorithm importantly. The algorithm combines the artificial immune theory and digital filter, selects property pretreatment method, designs and implements a suit of cold rolled strip surface defect segmentation algorithm, then tests a great deal of images collecting from produce present, analyses the segmentation performance of this method. At the end of the paper, several methods are compared to each other and draw a conclusion as following:1. Otsu can not segment the rolled steel surface defects. As the abnormal of the light, roughness of the background and defect size and so on, all of them will bring on the invalidation of the threshold, accordingly, the threshold can not segment the defects commendably.2. The single channel optimal filter has good result of segmentation, but it is difficult to check out when the inspection texture direction disagree with the training defect. The performance of the multi-channel optimized filters is not changed with the defects size, direction and background roughness. It is shown that multi channel optimal filter has strong ability. How to fuse each channels information to reach the best segmentation quality, most of the time depends on experience.3. The defect segmentation algorithm that this paper implements based on edge information can locate the defect accurately, it overcome the influence of factors such as the direction of the defect, light abnormal and so on. It has strong flexibility.4. The defect detection algorithm based on artificial immune theory can increase the energy mean rate between the defect and background. It can segment the defect simply and fast. This algorithm has real-time and strong ability. |
