Smart Vehicle Obstacle Detection by Camera and LiDAR Image Fusion

This paper proposes a novel obstacle detection method for autonomous vehicles that combines camera and LiDAR image fusion techniques.The proposed method employs the DeepLabV3+ algorithm with an attention mechanism for camera image segmentation and a centroid algorithm with scanning line bundle-based segmentation for LiDAR image processing.The processed images are then fused using the Local Non-Subsampled Shear Transform (LNSST) algorithm, which enhances the detail information and improves the recognition speed and accuracy.Experimental results demonstrate that the proposed method achieves superior performance in complex scenes, partially occluded objects, and long-range target detection compared to state-of-the-art algorithms.The proposed method significantly improves the environment perception capabilities of autonomous vehicles, contributing to safer and more efficient navigation in complex driving scenarios.