序列图像的目标识别算法改进

提出一种改进的动态模板匹配算法，将仿射不变矩作为提取目标时的不变特征，同时采用一种改进的canny算法对目标进行边缘检测，以去除伪边缘，利用长方形模板代替方形模板，提高算法的实时性。实验结果表明，该算法的识别率高于传统动态模板匹配算法，同时可以有效降低计算量。

Improvement of Target Identification Algorithm for Sequence Image

An improved dynamic template matching algorithm is proposed, which uses affine invariant moment as the invariable characteristic, when it recognizes goal. It uses one kind of improved canny algorithm to carry on edge detection to the goal, and removes false edge, and replaces the square shape template by using the rectangular template. This algorithm can improve real time. Experimental results indicate that, compared with traditional dynamic template matching algorithm, this algorithm has higher recognition rate and can down the computing quantity.