Lidar technology has been widely applied in various fields such as autonomous driving, robotics, and urban planning due to its ability to accurately capture three-dimensional environmental information. However, existing methods for processing large-scale lidar point cloud data suffer from issues such as low computational efficiency, ineffective aggregation of multi-scale and contextual information, and failure to meet real-time requirements. Therefore, how to improve the speed of processing large-scale LiDAR point cloud data while achieving accurate segmentation remains an important technical challenge. This paper presents a laser radar point cloud segmentation method based on view projection that addresses the aforementioned challenges and provides a practical solution for real-time, high-quality point cloud processing in large-scale applications. Firstly, the proposed method transforms 3D point cloud data into 2D images using projection techniques, reducing data complexity and improving computational efficiency to meet real-time requirements. As a result, it also enables effective aggregation of multi-scale and contextual information. Subsequently, the obtained image data is applied to mature 2D image processing methods for segmentation. Secondly, a pyramid feature aggregation module is introduced to effectively extract and utilize information from various scales. This module integrates multi-scale features and effectively compensates for potential information insufficiency of single-scale features, thereby improving the accuracy and robustness of point cloud segmentation. Furthermore, a deep multi-scale aggregation context module is designed to enhance the network’s ability to extract key contextual information from complex structures and scenes in the point cloud. To evaluate the effectiveness of the proposed method, we conducted comprehensive experiments on a widely covered LiDAR dataset and compared its performance against existing methods. The experimental results demonstrate that the method achieves outstanding performance in terms of accuracy, robustness, and scalability.