DTW算法的程序，申请两个n*m的矩阵D、d,分别为累积距离和帧匹配距离。这里n和m为测试模版与参考模版的帧数。然后通过一个循环计算两个模版的帧匹配距离d。接下来进行动态规划，为每个格点 （i，j）都计算其三个可能的前续格点的累积距离D1，D2，D3。考虑到边界问题，有些前续格点可能不存在，因此加入一些判断条件最后利用最小值函数min()，找到三个前续格点的累积 距离作为累积距离，与当前帧的匹配距离d（i，j）相加，作为当前格点的累积距离。该计算过程一直达到格点（n，m），并将D（n，m）输出，作为模版匹配的结果。-DTW algorithm procedures, apply two n* m matrix D, d, respectively, accumulated distance and fr a me matching distance. Where n and m is a reference test pattern and the template fr a mes. Then through a loop to calculate two fr a mes template matching distance d. Next, dynamic programming, for each grid point　(i, j) are calculated from the cumulative D1 before the renewal of its three possible lattice point, D2, D3. Taking into account the boundary problem, some former continued grid point may not exist, so add some conditions to determine the final use of the minimum function min (), found three former continued accumulation of grid points　cumulative distance as the distance from the current fr a me to match the d (i, j) added, and a current cumulative distance grid. This calculation has reached the grid (n, m), and D (n, m) output, as a result of template matching.