A Real-Time Vehicle Detection Algorithm Based on Sparse Point Clouds and Dempster-Shafer Fusion Theory*

摘要

In order to improve real-time performance and reduce the dependence on computing resources, we present a novel vehicle detection algorithm based on sparse point clouds in this paper. In point clouds segmentation, virtual laser line is proposed and our fast two-step segmentation method has proved to be time-efficient. Since accuracy and real-time capabilities are all crucial for autonomous vehicles, we utilize multiple features analysis and dempster-shafer(D-S) fusion theory to improve detection accuracy. Validation tests and experimental results show our method has a high performance in real urban traffic situations.