Speech can be represented as a time/frequency distribution of energy using a multi-band filter bank. A Markov random field model, which takes into account the possible time asynch...
This paper discusses theoretical and experimental aspects of gradient-based approaches to the direct optimization of policy performance in controlled ??? ?s. We introduce ??? ?, a...
This paper presents a direct and stochastic technique for real time estimation of on board camera position and orientation—the ego-motion problem. An on board stereo vision syste...
In this paper we present a new technique to extract layers in a video sequence. To this end, we assume that the observed scene is composed of several transparent layers, that their...
In this paper we propose a new probabilistic relaxation framework to perform robust multiple motion estimation and segmentation from a sequence of images. Our approach uses displa...