Abstract: A new defect detection algorithm base on Support Vector Data Description (SVDD) is proposed. A fabric texture model is built on the gray-level histogram of textural fabric image. Two Gray-level Co-occurrence Matrix (GLCM) features are used to characterize the fabric texture. And an adaptive quantization scheme base on the texture mode is proposed to reduce the size of GLCM and reduce the computational complexity of feature extraction. Besides, two new features are proposed to characterize the continuous property of the fabric defects. The SVDD classifier is used as a detector for defect detection. Experimental results of real fabric defects are provided to validate the effectiveness and robustness of the proposed detection algorithm. And a prototyped detection system is built to evaluate the real-time performance of the detection algorithm.