Normalized cross correlation image stitching algorithm based on minimum spanning tree

This paper presents a normalized cross correlation optimization algorithm based on minimum spanning tree. This method smoothes the original image and generates normalized correlation matrix for the smoothing image. According to the generated normalized cross-correlation matrix, the maximum value and corresponding index of each row and column (one point in each image relative to all corresponding points in another image) are obtained; according to the result, feature points are matched, and the minimum spanning tree is used in the matching process to plan the fastest traversal route. Two groups of pictures are used to calculate the feature point detection time, the matching time of feature points and the total matching time. Experiments show that the improved algorithm can save 2.72s and 0.53s, 2.68s and 0.56s compared with the traditional normalized cross-correlation algorithm and the optimized cross-correlation algorithm without minimum spanning tree. Experiments show that the proposed algorithm has higher accuracy and convenience, cost-effective and easy to implement.