An image matching algorithm based on FAST and sub-pixel accuracy

Image matching is the key technology of three-dimensional registration. Traditional image matching methods such as SIFT(DoG) and Harris are too computationally intensive for use in real-time applications of any complexity. A new image matching algorithm based on FAST corner detection and sub-pixel accuracy is presented which can reduce the computation of feature points detection, keep the high stability and meet the real-time requirement. Experiment results show that our method improved the speed of feature points matching with high stability.