The Research On Intelligent Recognition Of Cord Fabric Defects

The effective detection of the cord fabric defects has a great significance on improving product quality and reduceing the cost of product.The quality inspection and control is very important in the production process of the cord fabric.But the defects detection is the most important part of it.At present, the line testing of the domestic fabric production is basically done by artificial vision,and the cord fabric defects detection is also by this.In the manual testing process,it would create the false detection and undetected defects because the human vision often has a certain bias due to inherent defects in humans.The purpose of this research is to find a rapid approach of the cord fabric defects detection based on computer platform.The paper puts forward their own research methods based on the study and research of fabric defects automatic detection at home and abroad.First we introduce to build hardware platform of the detection system, and then is the defects automatic detection.It is included three parts,namely,image preprocessing,image segmentation and identification classification of the defects.In the part of image preprocessing,we first changed the original image into a grayscale image,and then compared the two kinds of de-noising filtering method,and selected the fast and efficient filtering method of median filtering, and then used Sobel operator to the image sharpening and reinforce.In the part of image segmentation,we first presented the fast method of the cord fabric defects detection based on the energy calculation,and then compared the effects of the three segmentation algorithms such as watershed algorithm and so on.And we obtained the conclusion that watershed algorithm is a relatively good segmentation algorithm,and segmented the image with the improved algorithm.In the part of identification and classification,we first finded five characteristic quantities according to the cord fabric defects characters,and the used the radial basis function network to identify and classify the defects.Through the experimental results,we found the research is valid and feasible.