Fabric Image Defect Segmentation Based On Grayscale LBP Co-occurrence Matrix And Spatial Weighted K-means

The textile industry has been one of the most important industries in China since ancient times.Under the background of global economic integration,how to make China's textile industry more competitive,fabric quality control is an important process.However,in the production process of the fabric,various defects will inevitably appear on the surface of the fabric.At present,most textile factories in China still rely on traditional manual detection in fabric defect detection,which has the disadvantages of low detection efficiency,high cost and harm to human health.With the development of digital images and machine vision,the digitization and automation of fabric defect detection has become possible.Replacing manual detection with computers can reduce costs and improve work efficiency.In this paper,the fabric image defect detection is taken as the research object,and the following three aspects are studied:(1)For the feature extraction of fabric images,based on the gray-level co-occurrence matrix and LBP features,the calculation method of the co-occurrence matrix is changed,and the two methods are combined,a feature extraction algorithm based on gray-level LBP co-occurrence matrix is proposed,which retains the advantages of the two methods and reduces the calculation amount.Meanwhile,interpolation is used to improve the error caused by distance selection.The experimental results show that the method is shorter in time and more suitable for real-time requirements.(2)For the image segmentation of the defect region,the k-means method is improved.Combined with the characteristics of the image data,the spatially weighted k-means algorithm is proposed to make it better applied to image segmentation.The experimental results show that the proposed method is more suitable for image segmentation and the segmentation result is better than the traditional method.(3)The design of the fabric image defect detection system was completed,including hardware selection and software implementation.This completes the construction of the hardware platform such as light source and illumination,image acquisition,and the implementation of the software host computer based on MFC and OpenCV.The system and the algorithm proposed in this paper are debugged and verified.The experimental results show that the platform basically meets the accuracy and real-time requirements of detection.Finally,the paper is summarized and further research is prospected.