论文部分内容阅读
在激光雷达目标识别中,目标姿态的精确估计可以有效地简化识别过程。现有的PDVA算法主要是针对地面结构化目标而提出的一种3D目标姿态估计方法。该方法利用模型坐标系(MCS)各个坐标轴的正方向向量来确定目标的三维姿态角,其有效性通过实验得到了验证。但该方法在确定MCS各坐标轴的正方向向量时,所消耗的时间比较多,影响了算法的执行效率。文中提出了一种改进的PDVA算法,利用聚类中心邻域判别CCND法来加速MCS各坐标轴的正方向向量的确定过程。采用四种地面军用车模型目标进行了仿真实验,实验结果显示,改进的PDVA算法的平均运行时间约占PDVA算法的66%,极大地提高了目标3D姿态估计的执行效率。
In laser radar target recognition, the accurate estimation of target attitude can effectively simplify the recognition process. The existing PDVA algorithm is aiming at the ground structure of the target proposed a 3D target attitude estimation method. The method uses the positive vector of each coordinate axis of the model coordinate system (MCS) to determine the three-dimensional attitude angle of the target, and its validity is verified experimentally. However, this method consumes more time in determining the positive vector of each coordinate axis of MCS, and affects the execution efficiency of the algorithm. In this paper, an improved PDVA algorithm is proposed, which uses the CCND method in the neighborhood of clustering center to accelerate the process of determining the positive direction vectors of MCS axes. Four ground military vehicle models are used to simulate the target. The experimental results show that the average run time of the improved PDVA algorithm accounts for about 66% of the PDVA algorithm, which greatly improves the execution efficiency of the target 3D attitude estimation.