Research on Centroid Tracking Algorithm for Oil Stain Defects on the Surface of Silk Cake

Abstract

As the basic material for the production of specific fabrics, the yarn quality of chemical fiber cake directly affects the fabric quality. The problem of oil on the surface of silk cake will lead to uneven yarn coloring, which will affect the quality of fabric. In order to solve this problem, a method of detecting oil stain defect of textile silk cake based on graph centroid tracking algorithm is proposed in this paper. In this method, the silk cake image is obtained, downsampled and gray-scale processed, and then Gaussian filtering and binarization are applied to enhance the image clarity. Next, the edge outline and its minimum external rectangle are drawn, and a mask image is constructed to highlight the oil area. By setting the oil pollution pixel range and processing the image after mask, log operator and expansion processing are used to enhance the oil pollution characteristics. Finally, the contours are extracted and the centroid coordinates are calculated. By comparing with the vertex coordinates of the minimum external rectangle, the oil pollution is automatically located. The method has high detection accuracy and short time, and can effectively solve the problem of oil stain defect detection of textile silk cake.