The Research Of Mesh Fabric Surface Quality Deetction System Based On Machitre Vision

With the development and maturity of machine vision technology, its application ismore extensive. Mesh fabric is being applied widely for its unique performance, novelappearance, and the performances improvement. Meanwhile, with the surface qualityrequirements higher and higher, the traditional manual testing methods have beenunable to meet the business needs.For the lag on the mesh fabric surface quality inspection, as well as domestic blankin the automatic detection of the surface quality of the mesh fabric, a mesh fabricsurface quality inspection system based on machine vision has been developed.According to the mesh fabric material properties, the hardware system was built;according to the mesh fabric texture properties, the image processing, feature extractionand defect detection methods were designed and the software platform was built, whichrealized the online real-time detection of surface quality of the mesh fabric.According to the noise generated in the image acquisition, combined with thestructural characteristics of the mesh fabric, taking into account the follow-uprecognition algorithm and real-time requirements, the mean filter is chose. In order toextract rapidly the regions of interest of images, achieve the robust machine visionsystem, a rapid edge detection method based on Sobel operator is proposed. Based onthe brightness of the image and analyzing the brightness curve, the adaptive thresholdsegmentation method based on brightness is proposed to eliminate the impact of theuneven illumination and effectively separate the background and foreground. Accordingto the texture characteristics of the mesh fabric, the defect segmentation method basedon texture features is proposed, which can segment the defects well. An imageprocessing method suitable for the mesh fabric is composed.By analyzing the mesh fabric defect categories and their characteristics, the featureset of mesh fabric are extracted, and a serial perceptron structure is designed, whichachieved the defect recognition and classification of mesh fabric.The experimental results show that this hardware architecture of detection systemdesigned is reasonable, software algorithms is efficient, which can achieve the onlinedetection of the mesh fabric. The inspection speed of online inspection system for meshfabric could up to60m/min and the defect recognition rate could up to95%or more,which meet the automation requirements of enterprises quite well. The successfulimplementation of the system has a great significance to enhance our mesh fabricautomatic detection level, as well as mesh fabric product quality.