基于目标的局部约束光流分析

针对LK方法在实际使用过程中存在计算量过于巨大,难以实时应用及计算误差偏大的问题.提出以一个对象的所有区域拥有相同的运动速度为局部约束,基于梯度选择有代表意义的特征位置代替全目标区域分析,使用一种自适应的基于运动范围的梯度算法,获得整个物体的运动特性.对实验结果的分析表明:该方法不但减少了局部约束光流法的运算量,而且在结果可靠性方面比典型的LK算法更佳.

Target-based Analysis on Optical Flow with Local Constraint

The Lucas and Kanada method is one of the major techniques to analyze the motion object in videos.However,in practical,the amount of calculation is too great to be used in the real-time work,and the calculation error is always too large.In this paper,the motion velocity of the whole target is obtained with the local constraint that is based on the case that one object only had one motion velocity,with the part of the object area which has a special grads character instead of the whole target,and with the gr...