基于相似度评估的激光雷达回波场景生成优化

Lidar Echo Scene Generation Optimization Based on Similarity Evaluation

基于三维场景构建和激光散射原理的激光雷达回波场景生成,可被用于模拟真实场景的激光回波特性,是激光阵列探测器性能测试的重要手段。为了高效地生成具有高仿真精度的动态场景,构建了由激光回波场景仿真相似度和场景计算耗时组成的采纳度并将其作为评价函数,实现了对激光回波场景空间网格划分数的优化。针对阵列单元为128×128的面阵探测器,利用所提方法对材质不同、距离不同的两种实验场景的激光回波的空间网格划分数进行了优化。当探测器单像元细分数为3×3,空间网格总划分数为384×384×L(L为距离精度决定的细分数)时,采纳度达到最大值。每帧场景的平均生成耗时均小于8 ms,实现了帧频为125 Hz的高帧频、高逼真度的动态场景生成。理论及实验结果均表明,在计算资源有限的情况下,所提方法可实现生成回波场景相似度和计算资源消耗间的平衡,是一种实用的三维激光回波场景生成优化方法。

Lidar echo scene generation, based on three-dimensional scene construction and laser scattering principle, can be used to simulate the laser echo characteristics of real scenes and is an important means for testing the performance of laser array detector. To efficiently generate dynamic scene with high simulation accuracy, an adoption degree composed of the simulation similarity of laser echo scene and scene calculation time is constructed and used as the evaluation function. The spatial mesh number of laser echo scenes is thereby optimized. For an area array detector with 128×128 elements, the proposed method is used to optimize the number of spatial meshes of laser echos in two experimental scenes with different materials and distances. When subdivision number of the detector single pixel is 3×3 and the total number of spatial meshes is 384×384×L(L is subdivision number determined by distance accuracy), the adoption degree reaches the maximum values. The average generation time of each frame scene is less than 8 ms, and the dynamic scene generation with high frame rate(125 Hz) and high fidelity is realized. The theoretical and experimental results show that this method can balance the similarity of the generated echo scene and computing resource consumption when the computing resources are limited, and it is a practical optimization method for three-dimensional laser echo scene generation.