An Image Mosaic Method for Defect Inspection of Steel Rotary Parts

According to the features of the inspection images for the steel rotary parts with defects, a novel image mosaic method, using Scale Invariant Feature Transform (SIFT) feature tracking with purifying feature points based on slope probability measure and RANSAC algorithm, is proposed. First, the method preprocesses the captured sequence images, and then implements projection transformation for these images. Then, the registration parameters for two adjacency images, using the SIFT algorithm and removal algorithm of the pseudo matching feature point pairs based on slope probability measure and RANSAC algorithm, can be solved to mosaic the defect inspection images of the parts with enough characteristic information. On this basis, a hardware-based method is used to perform image stitching of the measured parts. Experimental results show that the method can produce a large number of the correct matching feature point pairs, and can get a seamless, clear surface image of the parts, which will settle the foundation for automatic accurate inspection of the surface defects on metal parts.