Détection d’obstacles routiers par stéréovision : l’approche « v-disparité »

This paper presents an innovating real time method for detecting road obstacles by on board stereovision. After having exposed the stakes of the obstacles detection, having presented the state of the art and having identified the problems to be solved, the developed approach is described. It is based on the analysis of the “v-disparity” image, which provides a suitable representation of the geometric content of the road scene. The method is then evaluated with respect to the problems generally affecting stereovision approaches. The theoretical good properties of the “v-disparity” algorithm — genericity, effectiveness, robustness, and computational efficiency — are thus experimentally confirmed. Thanks to the performances obtained, the described algorithm is used for on board driving assistance applications.