Accurate multi-object perception system is the core component of autonomous driving technology. Aiming at the problem that single sensor is susceptible to environmental interference in complex traffic scenarios, resulting in missed detection and false detection, this paper proposes a multi-object detection algorithm based on decision-level fusion of LiDAR and camera. Firstly, temporal-spatial alignment is performed for LiDAR and camera to ensure the consistency of data in time and space domains. Secondly, PointPillars and YOLOv5 algorithms are adopted to detect objects on preprocessed point cloud data and image data respectively. Finally, Intersection over Union (IoU) matching, D-S evidence theory and weighted box fusion are utilized to realize decision-level fusion of detection results from two sensors. Experimental results demonstrate that the proposed fusion method achieves better detection accuracy than single sensors on both KITTI and nuScenes autonomous driving datasets, which effectively improves the accuracy and robustness of multi-object detection.