The Mean shift algorithm of moving target tracking accurately based on tower decomposition

When the object move fast or the frame rate of camera is low, the Mean shift tracker proposed may lost it. In order to overcome this disadvantage, an improved method is presented. Firstly we transform the image by the pyramid algorithm and computer Integral histogram of the low resolution image. Secondly, find the optical matching point as coarse object location through histogram match, then use Mean shift tracker to find the accurate matching point in the origin image. The experiments show that our method can achieve a better accuracy without losting real time ability when tracking fast motion object.