A Skew Detection Method of Fabric Images Based on Multi-threshold Analysis

The inevitable skew of the fabric images during scanning has brought great difficulties for the subsequent image processing operations. In the paper, a new skew detection method of fabric images based on the multi-threshold analysis is proposed: First, Sobel operator gradient is used to extract the weft information, and then improve the algorithm´s computational speed by introducing the aberration and more threshold values. To improve the accuracy, false extreme elimination, the mean weight operations and the error compensation are added when using hierarchical Hough transform to detect the skew angle of the weft. Thus, the skew angle of the fabric image is acquired finally. The results show that this method can detect the skew angle of fabric image fast, and the test results are accurate.