形态梯度恒常的复值小波光流求解

An estimation of complex wavelet optical flow with invariant morphological gradient

为提高光流估计的鲁棒性,假设运动物体形态梯度为不变量,据此提出了一种基于形态梯度恒常性的小波光流求解方法,以形态梯度恒常假设构建形态光流基本方程,并将其映射至小波域,利用小波的多尺度多分辨率特性提高光流求解精度,引入复值小波以降低相位震荡的影响.通过最小二乘法求解超定的小波光流方程组以解得光流矢量,并用中值滤波去除光流场中的异常分量.实验表明,提出的方法在光流求解精度及算法稳定性方面优于传统光流求解方法.

To improve the robustness of optical flow estimations, it is usually assumed that the morphological gradient of moving bodies is invariant. On that basis, a solving method for wavelet optical flow was proposed. The basic equations of morphological optical flow were developed. Then the equations were mapped to the wavelet field in order to improve the precision of optical flow estimation by using the multi- scale and multi-resolution features. Complex wavelet was introduced so as to reduce the effect of phase oscillation. Optical flow vectors were evaluated by solving the over-determined wavelet optical flow equations with least squares method, and the abnormal velocity components were removed through median filter. The results of experiments showed that this method is better than the classic image sequence optical flow method.