Texture Defect Inspection Method Using Difference Statistics Feature in Wavelet Domain for Human Computer Interface

In this paper, a texture surface defect detection method based on wavelet difference statistics is proposed. In this method, the wavelet function is firstly used to decompose the texture graph, and the image is divided into non-overlapping sub-windows to calculate the texture features. Mahalanobis is used to judge defects in this paper. Experimental results show that this method is convenient, fast and applicable.