In this paper, we present a method for segmenting 3D digital objects into discrete plane segments. This segmentation is the first step in obtaining a polyhedization of a discrete object with the property of reversibility. This constraint requires that the discretization result for polyhedrization be exactly the same as the initial set of points. One of our objectives is to reduce the number of planes in our segmentation, and achieve more efficient surface analysis algorithms. In 3D space, direction and starting point are common issues. Our method attempts to achieve segmentation by considering surfels one after the other and agglomerating them with their neighbors based on a distance ranking. This method enables the recognition of critical points on the boundary of a plane segment. A medical application is illustrated by the presentation of a tumor segmentation.