基于本征和非本征正则项的光流估计

Optical Flow Estimation Based on Intrinsic and Extrinsic Regularization

针对光流约束的孔径问题,结合本征和非本征两种正则化处理提出一种自适应光流估计方法。详细讨论了光流估计模型、Euler-Lagrange方程和时空导数的计算等问题。实验结果表明了这种方法的有效性,与经典的Horn-Schunck和Nagel-Enkelmann方法相比,能够更好地保持光流场的不连续边缘,同时在图像平滑或少纹理区域自适应地减弱光流约束作用。

To overcome the aperture problem of the optical flow constraint, an adaptive algorithm for optical flow estimation is proposed based on intrinsic and extrinsic regularization. The adaptive optical flow estimation model, Euler-Lagrange equation of this model and the spatial-temporal gradient are thoroughly studied in this paper. Experimental results demonstrate the effective of the proposed method. Compared with classical Horn-Schunck and Nagel-Enkelmann method, the edge incontinuity of optical flow field can be preserved much better, and in the area of texture scarcity, the optical flow constraint can be decreased adaptively.