Fabric Defect Detection Method And Its Software Design Based On DM642 Image Processing System

So far, fabric defect detection has still been a challenging problem for researchers. In this paper, block method, improved between-class method and method based on gray histogram are proposed by studying pixel gray-value characteristics of defect image and its corresponding defect-free image. The detection principles of the three methods are simple, and all detection time needed by three methods are quite short. The experimental results demonstrate that detection rates of three methods are relatively high.In this paper, fabric defect detection method based on Gabor filter masks is also proposed. In this method, directional selectivity, symmetry, frequency bandwidth selectivity and center frequency selectivity of Gabor filter are utilized. The parameters of odd symmetric optimal Gabor filter which can approximately describe a defect-free fabric image are optimized by GA. Then, even symmetric and odd symmetric optimal Gabor filter masks are derived. An fabric sample image and its corresponding defect-free image are filtered by using this two filter masks respectively in order to obtain two filtered sample images and two filtered defect-free images, and then we can get two binary images by conducting binarization operation on two filtered sample images respectively, in which the threshold values are obtained on the basis of the two filtered defect-free images separatively. The final segmentation result can be acquired by fusing the two binary images. The experimental results show that the detection rate of this method is much higher.In addition, software design of the fabric defect detection method based on gray histogram is accomplished based on the DM642 image processing system. The reading module of BMP files, image acquisition module, defect detection module, image preservation module and image display module of the software are designed successfully. The running results of the software show that defect detection results are the same as those obtained in MATLAB simulation environment.