图像匹配中KNN与RANSAC相结合的改进算法

Improved algorithm with combination of KNN and RANSAC in image matching

文章提出一种将KNN与RANSAC相结合的改进算法。通过获取最近邻与次近邻值并根据双向匹配原则,设计匹配不相关性的衡量因子,对KNN算法进行了改进;对RANSAC算法的代价函数和抽样规则进行了改进;最后将两种算法相结合,实现了速度快、自适应强,匹配精确的匹配算法。实验数据表明,该算法鲁棒性较强,自适应性较高,匹配速度较快。

This paper presents an improved algorithm combining KNN and RANSAC. The KNN algorithm is improved by obtaining the nearest neighbor and next nearest neighbor values and designing the measurement factor of match uncorrelation according to the two-way matching principle. The cost function and sampling rules of RANSAC algorithm are improved. Finally, the two algorithms are combined to realize a fast, adaptive and accurate matching algorithm. Experimental data show that the algorithm has strong robustness, high adaptability and fast matching speed.