Patterned fabric defect detection is a challenge in automatic inspecti on of fabric.In this paper,the mean-hash feature was introduced to represent the structure and gray features o f the patterned fabric,and realized the defects detection of patterned fabric.First,based on the periodic distance o f the patterned fabric,the mean-hash feature dictionary of the defect-free fabric was constructed.Next,the mean-ha sh feature and the gray feature in the testing fabric image blocks were extracted,and two difference maps were obtained .One difference map was obtained by matching the mean-hash feature dictionary using the Hamming distance,and the ot her one was obtained by comparing the gray feature with the average gray value of the entire image.Fina lly,two feature maps were summed and defect position was located.The experimental results with different patterne d fabrics show that the mean-hash feature is effectively used to represent the structural characteristics of periodic patt erns.And combining the gray features, hash algorithm improves the success rate of patterned fabric defect detection,a nd has good visual effects.