基于投影特征的快速图像匹配算法

Fast Image Matching Algorithm Based on Projection

针对传统的模板匹配算法存在的问题,提出了一种基于投影特征的快速匹配算法。该算法对图像进行投影得到一维投影特征。在搜索图投影计算的过程中,根据模板投影特征的极值,动态判定投影的灰度值,从而减少投影过程中非匹配区域的冗余像素的投影,运用KMP算法实现投影特征的快速匹配。在KMP快速匹配的基础上,进行NC归一化相关计算,得出精确的匹配位置。实验结果表明,该算法既能保证匹配精度,又提高了运算速度,是一种有效的算法。

A new algorithm of fast template matching based on projection is proposed after the problem of the algorithm of traditional template matching is analyzed. It projects the image to get 1D data. In the process of reference image project-ing, based on the value range of the data gotten af-ter template projection, real-time gray value judg-ment is realized during the projection, so that the redundant pixels of non - matched area. KMP - al-gorithm is used for pattern matching. Finally the NC （normalized correlation） is used to get the right position. The algorithm is shown as a robust algorithm with quick operation speed and fine pre-cision.