A Segmentation Algorithm for Irregular Three Dimensional Objects Fragments

Abstract:

The rule of dynamic average normal vector is proposed to segment irregular three dimensional object fragments by breadth first. Any triangle is chosen at random from geometric model of fragment as a seed mesh. Through the comparison, the triangle, whose normal vector is coincident to the seed mesh, will be added in. The average normal vector is calculated for the increased seed mesh and the comparing process is continued until the primary segmentation is obtained. The interrelated regions are classified and amalgamated to farther optimize the results of segmentation. Experimental results demonstrate the algorithm is efficient and robust.