Difficulties exist in automatically inspecting surface defects because of the low intensity image contrast. To overcome these difficulties, a textures analysis method for detecting defects on the magnetic tile surfaces was described. In this methodology the original image was divided into several equal sized squares, and decomposed based on a fast discrete curvelet transform (FDCT) at different scales and orientations. Then thel^2norms on the curvelet coefficients were calculated as the feature vector for support vector machine (SVM) classifier. The experimental results showed that the defects retrieval accuracy achieved 83% when defects accounted for more than 1/64 of magnetic tile image.
