📄 中文摘要
激光雷达传感器被广泛认为是自动驾驶的重要组成部分，但高分辨率传感器价格昂贵，而经济型低分辨率传感器产生的稀疏点云常常缺失关键细节。激光雷达超分辨率通过深度学习技术增强稀疏点云，解决了这一挑战，缩小了不同传感器类型之间的差距，并在实际应用中实现了跨传感器兼容性。该研究首次对自动驾驶领域的激光雷达超分辨率方法进行了全面的调查，尽管实际部署的重要性不容忽视，但迄今为止尚未进行系统性的综述。现有方法被组织为四个类别：基于卷积神经网络的架构、基于模型的深度展开、隐式方法等。
基于深度学习的激光雷达超分辨率在自动驾驶中的综合调查
📄 English Summary
A Comprehensive Survey on Deep Learning-Based LiDAR Super-Resolution for Autonomous Driving
LiDAR sensors are considered essential for autonomous driving, yet high-resolution sensors are expensive while affordable low-resolution sensors produce sparse point clouds that often miss critical details. LiDAR super-resolution leverages deep learning to enhance these sparse point clouds, addressing the challenge and bridging the gap between different sensor types, thus enabling cross-sensor compatibility in real-world applications. This study presents the first comprehensive survey of LiDAR super-resolution methods for autonomous driving. Despite the significance of practical deployment, no systematic review has been conducted until now. Existing approaches are categorized into four groups: CNN-based architectures, model-based deep unrolling, implicit methods, and others.