一种面向视频监控的三维复杂地形简化方法

A 3D complicated terrain simplification method oriented to video surveillance

有效的三维地形表示方法,在视频监控中科学合理选择部署摄像机位置,是提高监控效果的基础。传统的三维地形表示方法为了追求与真实地形的逼近,采样点密度高,且平坦区域与复杂区域均采用相同的采样密度,用于计算摄像机覆盖范围时会增加计算复杂度。本文结合地形特征与摄像机的特性,提出了一种面向视频监控的三维复杂地形简化算法。基于新的地形局部粗糙度,及相邻地形网格的夹角与摄像机的视场角间的关系,判断相邻网格是否可以合并。通过实验验证了本文算法的简化效果及有效性,对于实验中的某实际地形,地形网格数目简化明显;可保证监控到简化区域,就能监控到实际地形区域。

Effective representation method of 3D terrain is the foundation of choosing camera position to be deployed reasonably and improving the surveillance effect.Traditional 3D terrain representation methods focus on approximation to the real terrain,so the density of sampling points is high.Moreover,there is same sampling density for flat region and complicated region.The computing cost is too high if calculating the coverage rate of video surveillance based on traditional terrain representation methods.By combining local terrain features and characteristics of camera,a 3D complicated terrain simplification algorithm oriented to video surveillance is proposed.Neighbor regions are determined to be merged or not based on new local land surface roughness and relationships of angles between field of view of camera and local terrain rectangle.The simplification results and effectiveness are verified by experiment.For some real terrain,the number of rectangles is reduced greatly.And the real terrain can be monitored when the simplification terrain is monitored.