一种新型智能跳跃式模板匹配算法

New Intelligent Skip Method of Template-matching

提出了一种新型智能跳跃式模板匹配算法,它以图像的边缘点特征作为匹配的特征空间,以模板和搜索子图对应边缘点梯度幅值的累计差值作为相似性测度,以一种新型智能跳跃式搜索方法作为搜索策略。与传统的归一化相关测度法、序贯相似性检测法等穷举搜索的算法相比,可以大大地减少搜索位置,缩短搜索时间。实验表明,该算法在保证精度的前提下可以有效地提高图像匹配的速度。

A new intelligent skip method of template - matching introduced. It use the edge point pixel as feature space, and use the method of adding the differences of the pixels＇ gray value according to the corresponding position as similarity metric, finally use a new intelligent skip method as searching strategy. Compare with classical method SSDA and NNPROD, the new method can greatly reduce the complexity of the calculation and reduce search position. The experiment shows that the method greatly increase the matching velocity without the loss of matching accuracy.