1.　　视频数据读取，读取相邻两帧中的人眼最敏感的亮度数据。 2.　　用后一帧的亮度矩阵减去前一帧亮度矩阵，计算出两帧的帧差。 3.　　迭代计算出帧差中的噪声的均值和标准差。 4.　　根据均值和标准差滤除噪声，得到变化区域。 5.　　用数学形态学运算得到对象的最终模板。-1. Video data is read, read two adjacent eyes of the brightness of the most sensitive data. 2. After a brightness of the matrix before a less brightness matrix, calculated two frames worse. 3. Iterative Frame calculated the noise mean and the standard deviation. 4. According to the mean and standard deviation of filtering noise, changes to be regional. 5. Operators using mathematical morphology to be the ultimate target template.