The target detection technology and target tracking technology of road test LIDAR have important application value in the field of automatic driving and intelligent transportation system, especially in the face of complex scenes, it is difficult for vehicle-mounted LIDAR to realize the data acquisition of the whole scene, which makes the local sensing of road test LIDAR crucial. However, in practical applications, there are still some challenges in target detection, such as the accuracy and efficiency of background filtering. To solve these problems, this paper proposes a background filtering algorithm based on voxel number accumulation. This algorithm is divided into three parts, which are data collection, data preprocessing and background filtering. The intersection is selected for data collection, followed by data preprocessing to remove the background point cloud and ground points and record the background voxel information using the cumulative operation and voxel number accumulation. Next, the background voxel files are utilized for background filtering operations to remove irrelevant background information. Through experimental verification, the method can effectively improve the accuracy and efficiency of background filtering, avoid the occurrence of misclassification by manually filtering and updating the background voxel file, and provide accurate input data for subsequent target clustering and classification. The algorithm has important application value, which can improve the accuracy and efficiency of target detection and promote the development of autonomous driving and intelligent transportation systems. Future research can further explore and optimize the background filtering algorithm to improve the robustness and adaptability of target detection.