改进的PPF算法实现目标识别

With the development of lidar technology, target recognition and tracking based on point cloud are being studied widely.In this paper, In this paper, we improve the traditionalPPF algorithm and use it for target recognition.After obtaining point cloud,The target to be identified is lifted by point cloud preprocessing.According to the PPF, we calculate the point-to-point feature of the target to be identified, and search it in the constructed hash table, Use target probability instead of rotation matrix to complete target recognition. Through experiments in a simulation scenario, a graph of the target recognition rate versus distance is finally obtained, when the target is close to the scene,the recognition rate over 80%. Since the point clouds captured at different angles are subject to different occlusions, we select three angles, and draw the target recognition probability curve at the corresponding angles. The experimental results show that different angles affect the probability of target recognition, but not the total recognition rate.