The development of digital image feature detection and matching software

Digital image feature detection and matching is an important research content in the field of computer vision and pattern recognition. Inspired by Harris corner detector, the Harris correlation detector and Harris correlation descriptor (HCD) was studied and improved. In this paper, the scale adaptive Gaussian filter was introduced to optimize the descriptor and then the RANSAC algorithm was used to filter the matching result accurately. The experimental results show that the improved descriptor is invariant to image rotation, linear change of illumination and scale change, and its matching performance is independent of the type of feature points. According to the above research, a software of image feature detection and matching based on Open CV was designed based on VS2010 platform, which can clearly and intuitively show the process and results of feature detection and matching of different feature point detectors and descriptors.