Pedestrian Identifying and Tracking Algorithm for Point Cloud Data with Sparse Problem

Abstract

Abstract. In this study, we propose a method for identifying and tracking pedestrians from 3D point cloud data by combining the CVC method and an improved tracking algorithm. For pedestrian identification, the previous CVC method is used, and for tracking, the tracking algorithm improved in this study is used to solve the density problem of point clouds and the separation problem of pedestrians in close proximity. We confirmed that the proposed method can detect the actual number of pedestrians more accurately than existing methods, and can also sufficiently solve the problem of occlusion.