车祸再现中三维车辆模型准确检索方法研究

TheTraffic Accident Reconstructions of 3D Model Retrieval Methods Accurate Vehicles

研究交通事故再现中根据三维模型的车辆检索问题。为解决事故责任的确定,需要根据视频记录通过车辆三维模型的检索结果再现交通事故过程。交通事故发生过程中,车辆因事故碰撞发生车体旋转,使得车辆三维特征不明显,传统的基于骨架特征的三维车辆检索方法由于车体的旋转,不能有效获取车辆特征,造成车辆三维模型检索的准确率不高。为解决此难题,提出基于同心球特征提取的三维车辆模型检索方法,选取车辆的质心作为模型球心,依据最大半径均分原则得到多个同心球,选取与球心最远的凸点为特征点,这样选取的特征点与空间坐标位置无关,避免了车体旋转对特征点的影响,通过匹配完成检索。实验表明,同心球特征提取方法能够提高三维车辆模型的检索准确性。

Research the traffic accident reconstructions of 3 d models according to the vehicle problem solving ac-cident liability for the retrieval of certain, according to the video record through the vehicle to 3 d model retrieval re-sults of reappearance accidents process the traffic accident process, the car because accident collision body rotation, allowing vehicle 3 d characteristics not obvious, based on the characteristics of the traditional framework of 3 d vehi-cles due to the rotation of the body retrieval method, can＇t acquire effectively vehicles features, cause vehicles 3 d model retrieval accuracy is not high. In order to solve the problem, based on the concentric ball feature extraction of 3 d vehicle model retrieval method, the selection of the vehicle as a model at its center, according to the maximum ra- dius get multiple concentric another principle ball, the selection and at its farthest protruding point for the characteris-tic and selected feature point with this space coordinate location has nothing to do, to avoid the body to the influence of the rotating feature points, through the matching complete retrieval experimental results show that this method can improve the 3 d model retrieval accuracy of vehicle.