Abstract:

A new approach based on improved Gaussian Mixture Model was developed to detect the defect on print fabric. Considering the detection accuracy is not high by using the traditional Gaussian Mixture Model based method, after analysis the characteristic of the print fabric image, this paper proposed a defect detection algorithm with adaptive partition block model. Experimental results indicate that by using our proposed method, the detection success rate can reach 94%. Furthermore, the proposed method can tackle with the illumination change and noise problem effectively, it is suitable for practical application.

Key words: print fabric, defect, Gaussian mixture model, partition block modeling