针对基于灰度相关的图像匹配方法速度慢的特点,提出了一种基于像素连通特征的快速图像匹配算法。算法先利用一定的像素分类原则,将所有像素分为5类,再按照一定的连接原则按列统计每个模板列长度的连通分量个数。用模板各列的连通分量个数与待搜索图的各统计值进行串模式匹配。分析及实验结果表明,算法在保持较高匹配精度的同时,运算速度显著提高。

The image matching algorithm based on gray correlation is usually slow in calculation,thus we put forward a fast image matching algorithm based on pixel connecting information. In this algorithm,the pixels were divided into five classes according to given classification principles. Then the numbers of connective parts in length of each models column were counted according to linking principle. The connective parts of each column of the model were matched with the statistical values of the image to be searched. The analysis and experiment result showed that this method can increase matching velocity markedly while keeping high matching accuracy.