基于像素字符串比较的图像快速匹配算法

在分析传统的模板匹配算法的基础上提出了一种新的基于字符串匹配的快速匹配算法。算法的思路是在模板图像上任意确定一列像素，并将这一列像素的灰度值看成是一个字符串，以此对原图像的每一列进行字符串匹配。如果在原图像上的某一列上找到了完全匹配的串，或者找到最大匹配的串，就找到了所要匹配的模板在图像中的可能位置。然后在所有找到的位置上再做进一步的字符串匹配。如此继续就可以确定模板图像在待匹配图像上的位置。算法在统计意义上保证了匹配效果，且提高了匹配速度。实验结果表明该算法是一种有效的图像匹配算法。

An Improved Fast Template Matching Algorithm Based on String Comparison

After analyzing conventional template matching methods,a new matching algorithm based on the idea of sting comparing is proposed. The algorithm first chooses a column (or a row) on the template image as a string,then uses it to compare each column (or row) on the original image. Whenever a complete matched string or best match string is found, the possible location of template on the original image is found. Then from the location obtained do similar comparison again. The algorithm continues until a desirable match is fulfilled.The algorithm ensures the matching effect in statistical sense,and improves the matching speed.The experiment shows the efficiency of the algorithm.