The visual inspection system was developed for defects detection on leather surfaces, which is an important component of automatic CAD/CAM cutting systems. The main functions of the system are quality control and raw material cutting. An efficient algorithm, which combines multiresolution approach, energy and entropy matrices, is presented for detection of defects embedded in leather surface images. A wavelet band selection procedure was developed to automatically determine the number of resolution levels and decompose subimages for the best discrimination of defects and removals of repetitive texture patterns in the image. An adaptive binary thresholding is then used to separate the defective regions from the uniform gray-level background in the restored image. The proposed methodology is able to efficiently detect several types of defects that current approaches cannot detect, and is fast enough to be used for real-time leather inspection.