Novel Image Registration Based on Harris Multi-Scale Corner Detection Algorithm

The performance of feature detector determines the precision of registration directly.In this paper,the multi-resolution idea is introduced into the classical Harris algorithm,and the wavelet-based formula for measuring the image intensity variation is developed,meanwhile,the auto-correlation matrix is obtained that reflected the scale variation information.Then,a novel Harris multi-scale corner detection algorithm is presented,which might overcome the drawback that the single-scale Harris detector usually leads to either missing significant corners or detecting false corners due to noise.In order to compensate for the orientation difference between two target images,a so-called "angle histogram" is calculated.Based on the rotation angle and feature sub-images,the Corner Pair Alignment Metric(CPAM) is defined.Finally,each corner matching pair is determined accurately using maximization of the AM.Experiments demonstrate that this algorithm has a good performance of accuracy,efficiency and robustness.