基于量子遗传算法的高效匹配搜索策略

An Efficient Search Strategy for Image Matching Based on Quantum Genetic Algorithms

在大规模源图像上进行图像匹配时,最佳匹配点的搜索策略是匹配算法时间性能的决定因素,设计高效匹配搜索策略是提高算法性能的关键。为了减少搜索时间和提高匹配实时性,本文基于匹配源图像划分和量子遗传算法基本原理,提出了面向大规模源图像匹配的目标淘汰搜索策略TESS。TESS将基于整幅源图像的全空间随机搜索的过程变成基于各个子图像的子空间并行搜索和逐步淘汰的过程,实现了匹配区域粗定位与匹配点精搜索的有效结合,从而大大缩短了最佳匹配点的搜索时间。实验结果表明,TESS搜索策略带来了匹配速度的极大提高,且时间加速比随匹配源图像规模的增大而增大。

When the source image scale is large enough,the search time will be excessively long if the entire space searching on the source image is carried on.In order to decrease the search time and increase the real-time matching capability,a target elimination search strategy（TESS） is proposed based on the source image division and the basic principle of QGA.Through changing the random searching process in the entire space to the parallel searching in each subspace and the subspaces gradually,TESS combines the matching region rough-estimating with the optimal matching point fine-searching effectively,so the search time decreases enormously.The experimental results indicate that TESS makes the matching efficiency is improved enormously and the time speedup increases with the source image scale increase.