Abstract：In the process of detecting the surface defect of a steel plate based on vision,images obtained by a detection system appear local partial light,increasing the difficulty of defect detection,due to the influence of factors such as illumination. According to the problem above-mentioned,an image enhancement method of dynamic average is proposed,making the situation that images have been the local partial light to improve greatly,and providing great convenience for the image segmentation. After image enhancing,image segmentation experiments have been done by using the method of maximum entropy segmentation. The results of experiment show that this method makes the steel plate surface defect detection more effective. It also verifies the feasibility of this method in the steel plate surface defect detection and its application value.

Keyword：defect detection;dynamic mean;the maximum entropy segmentation;image enhancement;