A Video Surveillance Oriented 3D Terrain Simplification Algorithm

In video surveillance fields, the effectiveness of 3D terrain description method is the base of improving monitoring effect, selecting positions of cameras to be deployed reasonably and scientifically. In this paper, a new local land surface roughness representation method is introduced, and then combined with features of camera forms the measurement of merging neighbouring regions. Experiments show the results of the proposed 3D terrain simplification method, and the effectiveness is also proved. For real 3D terrain data used in the experiments, the number of grids is reduced sharply. The difference between coverage rate to the source terrain and the simplification terrain is small, and has almost the same coverage rate when the number of cameras is enough, which means that covering the source terrain can be guaranteed by covering the simplification terrain.