一种改进的Mean Shift跟踪算法

在复杂背景的图像中,用直方图作为目标的特征模板,依据颜色分布进行匹配,具有较好的稳定性.Mean Shift算法是计算最优解的一个高精度算法,能在良好的目标初始化的前提下跟踪到无遮掩的目标.但其新目标由手工标定,特征模板计算量很大,且容易丢失遮掩情况下的目标,所以对Mean Shift算法进行了四处改进.改进后的算法能够准确地初始化、并快速精确地跟踪目标.

A promotion Means Shift tracking algorithm

Histogram as the characteristic template of target and matched based on color distribution has good stability in the complex background image. Mean Shift algorithm is a high-precision algorithm to calculate the optimal solution, can track unobstructed target on the premise of that target is initialized well. But the new target is manual calibration, the characteristic template will lead to huge calculation, and easy to lose the obstructed target. Therefore, carry out four to improve the Mean Shift algorithm. The improved algorithm can initialize the target accurately, and track the target quickly and accurately.