An Improved Relaxation Algorithm for Image Matching

This paper presents an improved relaxation algorithm for image matching. The original relaxation algorithm includes two steps: rough matching and accurate matching. It usually is time-consuming for most practical problems. With the improved relaxation algorithm proposed in this paper, the number of matching points that take part in the next pass iteration can be reduced, consequently, the algorithm has the higher efficiency. An image stabilization system is taken as a platform to verify the algorithm. The experimental results demonstrate the significant improvement of the iteration speed.