Abstract:The method for recognizing the photoelectric fabric weft has the problems of low accuracy, poor adaptability and long time. A method of fabric weft skew detection for machine vision based on Sobel edge direction and progressive probabilistic Hough transform (PPHT) is proposed, which can quickly and accurately recognize the angle of fabric weft skew. Firstly, the fabric image collected by the industrial camera is subjected to frequency domain filtering and inverse transformation to enhance the weft lines in the fabric image; then the Sobel direction operator is used to convolve the fabric image to obtain the edge direction pattern of the fabric image, and thresholding segments the weft lines region in the edge direction pattern; finally, the weft lines region is skeletonized, and the cumulative probability Hough transform is used to detect the lines skew angle of the skeleton of the approximate weft lines region, that is, the fabric weft skew. The weft skew detection test was carried out on fabric samples of different types. The test result was that the deviation value of the fabric weft skew was less than 0.1°, and the detection time was less than 0.6 s. The results prove that the proposed method can meet the requirements of accuracy, real-time and versatility of the machines vision weft-straightener.
