Penggunaan Gray Level Co-Occurance Matrix Dari Koefisien Aproksimasi Wavelet untuk Deteksi Cacat Tekstil

Textile defect detection is still carried out manually and it is hard to detect textile defect more than 60%. Therefore, this research aims for automatic textile defect detection by using Gray Level Co-occurance Matrix (GLCM) of wavelet approximation coefficients and evaluating the performance of this approach. The textile images were tiled into eight parts, to get clearer texture defects. Each part is decomposed into two levels wavelet coefficient. GLCM calculated from the wavelet approximation coefficient level one and level two is the feature used. Each set is conducted by using one image of non-defect textile and two images of two types of defect textile. Euclidean distances between each part of non-defective image and all parts in the first and the second defective image are calculated. From the simulations, it turns out that the GLCM of wavelet approximation coefficient level two is able to detect more the ±70% of the existing defects