摘要

In order to improve the accuracy of optical flow computation, a variational partial differential based multi-model optical flow calculation method is proposed. Based on the variational partial differential optical flow calculation framework, the multi-data term model and smoothness term model are fused to estimate optical flow. Multi-channel color information and gradient information are employed to construct the data term. The global smoothness constraint model is employed to construct the smoothness term. The robust penalty function is employed to penalize the data term and smoothness term so that the outlier influences can be removed. To solve the problem of large displacement, the improved median filter pyramid model is employed to estimate optical flow under multi-resolution framework. The experiments on synthetic and real data demonstrate that the proposed method features high accuracy and fast calculation speed, and is an optical flow estimation method with good comprehensive performance.

全文