Trends and Techniques in 3D Reconstruction and Rendering: A Survey with Emphasis on Gaussian Splatting

Area of Science:

• Computer Graphics
• 3D Vision
• Robotics
• Virtual and Augmented Reality

Background:

• Three-Dimensional Gaussian Splatting (3DGS) is a novel technique for rapid novel view synthesis.
• It enables real-time 3D shape estimation without relying on neural networks.
• 3DGS has gained significant traction across diverse fields like robotics and AR/VR.

Purpose of the Study:

• To provide a comprehensive systematic review of Three-Dimensional Gaussian Splatting (3DGS).
• To outline the historical development and recent advancements in 3D reconstruction and rendering.
• To identify current challenges and suggest future research avenues in 3D vision.

Main Methods:

• Literature survey focusing on 3D reconstruction and rendering techniques.
• Analysis of foundational and innovation-driven perspectives in 3DGS.
• Summary of common datasets and evaluation metrics used in the field.

Main Results:

• 3DGS offers accelerated rendering for novel view synthesis.
• The technique facilitates real-time 3D shape estimation, crucial for various applications.
• A review of existing literature, datasets, and metrics provides a foundational understanding of 3DGS.

Conclusions:

• 3DGS represents a significant advancement in real-time 3D vision and graphics.
• Further research is needed to address current challenges and unlock the full potential of 3DGS.
• This survey serves as a valuable resource for researchers in 3D reconstruction and rendering.