Point cloud is a highly sought-after form of 3D data representation that has ignited a surge in multimedia research over the years. Given its ability to accurately capture the geometry of the real world, point clouds have become indispensable for various applications such as immersive media and 3D reconstruction. As such, it is imperative to develop advanced point cloud processing algorithms that can meet the ever-growing demands of these applications. This chapter provides a comprehensive overview of deep learning-based point cloud processing technologies. This chapter begins with an introduction to 3D world representation, then turns to the 3D reconstruction based on point clouds, covering the definition, properties, categorization, and different acquisition techniques. Subsequently, the discussion delves into the benchmark datasets commonly employed by the research community to enhance the processing and understanding of point cloud data. The discussion then shifts the focus toward various 3D perception tasks involved in point cloud processing and their application fields. This chapter equips readers with the necessary foundations to delve deeper into the subsequent chapters of the book.