AUTOMATIC SURFACE CRACK DETECTION ON TEXTURED MATERIALS

A novel image transformation is proposed for the automatic detection o f crack-like faults on inspected surfaces with ''busy'' or patterned b ackgrounds (textured surfaces). The textured background is modeled by the proposed transformation, which consists of the Walsh transformatio n of a nonlinear function of the image intensity. The algorithm includ es three stages: training the system to learn the underlying backgroun d textural pattern by using faultless images, obtaining the local diff erence from the underlying texture for the images to be tested, and po stprocessing the difference map to isolate the fault pixels. By this m ethod, crack-like faults in random and regularly textured backgrounds can be detected reliably and efficiently. The performance of the algor ithm is demonstrated on artificially created faults on some Brodatz te xture images as well as on some real images of materials with genuine faults on them.