An Improved Algorithm for Gray Image Matching

摘要

During the process of analyzing the image matching principle, we find that reducing the size of template images' searching space in reference images and reducing the correlation calculation between template images and sub-images can reduce the total amount of calculation of the matching process. A new matching algorithm is designed for the image matching process. The reference image is preprocessed first; then sample the preprocessed reference image with the normalized gray value as probability density; next the rough matching points will be obtained by selecting proper sets of sub-images and updating threshold dynamically; and lastly the process of fine matching is executed. Least-squares method is used to obtain the stationary point of quadratic curve in the nine points' neighborhood with matching points as center, which is the exact matching point eventually. By this way, the complexity of the new algorithm is reduced. The experimental results show that the new algorithm has certain superiority and practicability.