A Fast Mutual Information Image Matching Method Based on Statistical Correlation

Abstract

The invention discloses a fast mutual information image matching method based on statistical correlation. First, the traditional method is used to calculate the respective histograms and entropy of the measured image and the first position reference subimage, and the joint histogram, joint entropy, and mutual information of the two images. Then, during the traversal search process, using the correlation between adjacent reference subgraphs, the histogram, entropy and joint entropy of the previous position reference subgraph are used as the benchmark, and the current value is calculated by the difference method The histogram, entropy and the joint entropy of the position reference subgraph and the actual measurement map, and calculate the mutual information; finally, find out the position with the maximum mutual information value in all search positions as the final matching and positioning result; In addition to a position reference sub-graph, the histogram, entropy and joint entropy of other position sub-graphs are all sequentially obtained based on the difference method, which improves the speed of mutual information matching while ensuring the matching accuracy; in addition, it can also be compared with existing methods. Combine to speed up matching in multiple ways.

Description

technical field [0001] The invention belongs to the technical fields of image matching and image positioning, and in particular relates to a fast mutual information image matching method based on statistical correlation. Background technique [0002] Image matching technology is a new technology developed on the basis of aerospace technology, satellite application technology, sensor technology, computer technology, image processing and pattern recognition. It has important application value in military fields such as terminal guidance, image target search and tracking. At present, matching technology is widely used in industrial control, medical image processing, remote sensing image processing, cartography, target recognition and other fields. In applications such as vision-based aircraft navigation and guidance, remote sensing satellite disaster monitoring and environmental monitoring, and medical image analysis, it is often necessary to match heterogeneous images acquire...

AI Technical Summary

Problems solved by technology

Among the above methods, the grayscale compression method and the feature extraction method will lose part of the grayscale information of the image and affect the matching accuracy, while the particle swarm optimization method is easy to fall into local extremum and affect the matching accuracy. The Powell method has higher requirements for the initial point. The selection of the initial point will directly affect the final search for the optimal value

In short, most of the traditional fast mutual information matching algorithms speed up the matching speed by reducing the grayscale or reducing the number of pixels participating in the mutual information measurement calculation, which will affect the matching accuracy to a certain extent.

Images