Template matching parallel implementation method and apparatus by combining large template graph

Abstract

The present invention discloses a template matching parallel implementation method and apparatus by combining a large template graph, and belongs to the field of normalized correlation graph templatematching. The normalized correlation coefficient calculation module logical framework comprises a template image gray value summation module, a template gray value square summation module, a real-timegraph gray value summation module, a real-time graph gray value square summation module, a real-time graph template gray value product summation module, a subsequent calculation module, a real-time graph template gray value total buffer module, a real-time graph gray value total buffer module, a real-time graph gray value total square buffer module, and other modules. According to the technical scheme of the present invention, by adding the real-time graph gray value total buffer module, the real-time graph gray value square total buffer module and the real-time graph template gray value product total buffer module, and by controlling the number of parallel channels implemented by comparing the sizes of the template graphs in the process, control of the logic of the big template graph mode and the small template graph mode is respectively executed, so that the structure is realized to be no longer limited by the number of parallel channels and the resource utilization is improved.

Description

technical field [0001] The invention relates to the field of normalized correlation image template matching, in particular to a matching parallel implementation method and device for fusing large template images. Background technique [0002] Template matching performs target localization by calculating the similarity measure between the template map representing the target and all positions in the search area in the real-time map, and the target has similar size and image as the template map. Because of its invariance to changes in brightness and contrast, the normalized correlation coefficient is a widely used measure for template matching. Assume that the real-time image and the template image are denoted by A and B respectively, and their dimensions are K×L and M×N pixels respectively. At any search position (u, v), 0≤u≤K-M, 0≤v≤L-N, the normalized correlation coefficient (NCC-Normalized Cross-Correlation) is defined as: [0003] [0004] where ∑∑ means is the m...

AI Technical Summary

Problems solved by technology

[0008] The technical problem to be solved by the present invention is to propose a matching parallel implementation method and device for merging large template graphs in view of the defects of the background technology, and realize how to provide a small template graph to realize the calculation of normalized correlation coefficients that adapt to large template graphs at the same time method, which can solve the problem of matching the size of the template graph in different stages of the above task

Images