International audienceThis paper evaluates rasterization strategies and the benefit of hierarchical representations (in particular attribute profiles) to classify point clouds. When such data comes from LiDAR acquisitions, a rasterization process if often performed to build an elevation map (possibly used together with multi or hyperspectral images). While some works use attribute profiles onsuch elevation data, we rather focus here on several LiDAR features rasterized and on their multilevel description to produceaccurate land cover maps over urban areas. Our experimental results obtained on LiDAR data from the university of Houstonindicate good classification results based on our rasters.Cet article traite de rastérisation par représentat...