Optimal point correspondence for image registration in 2D images

The conventional search methods have computational complexity problem and imprecision problem in correspondence matching process. To resolve these problems, we propose how to effectively make feature space (distance map) and how to rapidly search the optimal point correspondence. The proposed distance map named Voronoi distance map is a 2-Dimensional surface that contains the distance information between each element(x-y coordinates) of image and the nearest feature point. The proposed distance map is efficiently created based on the priority-based calculation algorithm. The general distance calculation algorithm has a time complexity of O(w * h * n). (w = width, h = height, n = the number of feature points) whereas the priority-based distance calculation algorithm is a effective method with a computational cost of O(w * h * logn). Also, The partition search algorithm is a efficient method that can detect corresponding points very rapidly because this method can reduce the search range by a quarter at a time. Experimental results show that the proposed method outperforms conventional methods in reducing computation time and detecting the optimal correspondence.