Recent progress for alpha matting and visual correspondence

This talk addresses two computer vision problems that have been extensively studied over the last decades, namely alpha matting and visual correspondence search. Alpha matting aims to extract a foreground object from a single natural image by recovering the partial transparency (alpha matte) of the foreground. I will present approaches to three fundamental challenges in interactive image matting: (i) Providing a fast and intuitive user interface; (ii) finding a good cost function for matting; and (iii) providing a benchmark that allows a quantitative comparison of matting results.

In the second part of the talk, I will highlight our recent work on visual correspondence search (i.e. stereo matching and optical flow). Here, the focus is on efficient local methods to achieve (i) disparity maps in real-time that are competitive with the state-of-the-art, and (ii) optical flow fields with very fine structures as well as large displacements.