A Review Of The Latest Research Technologies Related To 3D Point Cloud

Abstract

In recent years, point clouds have been widely used in fields such as computer vision, medical image processing, virtual and augmented reality, autonomous driving, and robotics. Despite the remarkable achievements of deep learning methods in processing 2D data, they still face some unique challenges when processing 3D point cloud data [1]. The unstructured and irregular nature of point clouds makes it difficult to directly apply traditional deep learning methods, so point cloud deep learning is still in its infancy. However, some progress has been made in the field of deep learning for point clouds. Researchers have proposed many innovative methods and network architectures for solving tasks such as classification, segmentation, generation, and detection of point cloud data. These methods include the network structure of PointNet [2], PointRCNN [9] and so on as well as various data enhancement and optimization strategies. These research results laid the foundation for the development of point cloud deep learning, and provided important reference and inspiration for future research.