Qualitative scene interpretation using planar surfaces

Our aim is to provide an autonomous vehicle moving into an indoor environme nt with a visual system to perform a qualitative 3D structure reconstructio n of the surrounding environment by recovering the different planar surface s present in the observed scene. The method is based on qualitative detection of planar surfaces by using pr ojective invariant constraints without the use of depth estimates. The goal is achieved by analyzing two images acquired by observing the scene from t wo different points of view. The method can be applied to both stereo image s and motion images. Our method recovers planar surfaces by clustering high variance interest po ints whose cross ratio measurements are preserved in two different perspect ive projections. Once interest points are extracted from each image, the cl ustering process requires to grouping corresponding points by preserving th e cross ratio measurements. We solve the twofold problem of finding corresponding points and grouping t he coplanar ones through a global optimization approach based on matching o f high relational graphs and clustering on the corresponding association gr aph through a relaxation labeling algorithm. Through our experimental tests, we found the method to be very fast to conv erge to a solution, showing how higher order interactions, instead to givin g rise to a more complex problem, help to speed-up the optimization process and to reach at same time good results.