Efficient Stereo and Optical Flow with Robust Similarity Measures

In this paper we address the problem of dense stereo matching and computation of optical flow. We propose a generalized dense correspondence computation algorithm, so that stereo matching and optical flow can be performed robustly and efficiently at the same time. We particularly target automotive applications and tested our method on real sequences from cameras mounted on vehicles. We performed an extensive evaluation of our method using different similarity measures and focused mainly on difficult real-world sequences with abrupt exposure changes. We did also evaluations on Middlebury data sets and provide many qualitative results on real images, some of which are provided by the adverse vision conditions challenge of the conference.