An energy-based joint motion and disparity estimation algorithm with an anisotropic diffusion operator is proposed to yield correct and dense displacement vectors. We propose two energy models; the joint estimation model and the simultaneous joint estimation model. In the joint estimation model, we compute the initial disparity in the current frame with the joint estimation constraint, using the left and right motions and the disparity in the previous frame; therefore, the model is prevented from being trapped in the local minima. Then, we regularize this disparity by using our proposed energy model. In the simultaneous joint estimation model, we propose an energy functional that consists of fidelity and smoothing terms for the left and right motions and the joint data terms. We estimate the left and right motions simultaneously in order to increase correctness. We use the Euler-Lagrange equation with variational methods and solve the equation with the finite difference method (FDM) to minimize the energy model. Experimental results show that the proposed algorithm provides accurate motion-disparity maps that reflect the constraints of motion and disparity, and preserve the discontinuities of the object boundaries well.
All Science Journal Classification (ASJC) codes
- Signal Processing
- Computer Vision and Pattern Recognition
- Electrical and Electronic Engineering