I. Introduction
Patterned fabric defect inspection is of vital importance in the quality control process. Currently, automated inspection of the defects based on computer vision has the advantages of high precision and efficiency over the human visual inspection which suffers from high error rate, high labor cost and slow inspection speed especially in the field of patterned fabrics. There are three main categories for the detection techniques of patterned fabric defects: statistical, spectral, and model based [1]. The statistical approaches provide various statistical data about the spatial relations among pixels. These approaches can be divided into first order, second order and higher order levels. Second order statistics approaches are widely used as it is most appropriate to human's vision system [1]. These approaches include the gray level co-occurrence matrix (GLCM), which has been proven to be a very powerful tool for texture analysis as it estimates the gray-level dependencies in a local neighborhood for a given pixel displacement and orientation[2]. Haralick et al. [3] derive many features extracted from GLCM, which are used mostly in texture analysis applications [4], [5], but in this paper, we intend to use information from GLCM itself rather than its features.