Estimation of 3D Motion from Stereo Images – Differential and Discrete Formulations

In this paper we analyze the problem of motion estimation from a sequence of stereo images. We formulate both the differential and discrete approaches of two methods. The differential approach uses differential optical flow whereas the discrete approaches uses feature correspondences. We use both methods to compute, first, the 3D velocity in the Z direction and, second, the complete motion parameters. The methods were extensively tested using synthetic images as well as real images and several conclusions are drawn from the results. We point out the critical factors for the methods. The real images are used without any illumination control of the scene in order to analyze the behavior of the methods in strongly noisy environments and low resolution depth maps.