ANN and ICA sparse code shrinkage de-noising based defect detection in pavement tiles

This paper is concerned with the problem of image analysis based detection of local defects embedded in pavement tiles surfaces. The technique developed is based on the ICA sparse code shrinkage denoising, the local 2D discrete Walsh transform and ANN. To reduce random noise, the ICA sparse code shrinkage de-noising is applied. Next, robust local features characterizing the surface texture are extracted based on the 2D Walsh transform and then analyzed by an artificial Neural Network. A 100% correct classification rate was obtained when testing the technique proposed on a set of surface images recorded from 400 tiles.