Three-Dimensional High Efficiency Video Coding

Description

3D-HEVC is a video coding standard developed as an extension to High Efficiency Video Coding (HEVC/H.265) specifically designed for three-dimensional video content. The technology addresses the unique requirements of stereoscopic 3D and multi-view video by introducing specialized coding tools that efficiently compress both texture (color) information and depth maps. Unlike conventional 2D video coding, 3D-HEVC must handle multiple correlated views of the same scene, requiring sophisticated prediction mechanisms to exploit inter-view redundancy while maintaining the ability to reconstruct intermediate views at the decoder.

The architecture of 3D-HEVC builds upon the HEVC framework but introduces several key extensions. The system supports multiple texture views along with corresponding depth maps that provide per-pixel depth information. Depth maps are typically represented as grayscale images where pixel intensity corresponds to distance from the camera. 3D-HEVC employs advanced prediction techniques including disparity-compensated prediction (DCP) that allows prediction between different views, similar to motion compensation between frames in temporal prediction. This inter-view prediction significantly reduces redundancy between adjacent camera views.

Key technical components include depth modeling modes (DMMs) specifically designed for depth map coding, which exploit the piecewise smooth nature of depth information. The standard supports various 3D video formats including stereo video (two views), multi-view video (multiple camera views), and multi-view video plus depth (MVD) formats. The coding process involves joint optimization of texture and depth coding, where rate-distortion optimization considers both the quality of reconstructed texture views and the accuracy of synthesized intermediate views.

In 3GPP networks, 3D-HEVC enables efficient delivery of immersive video services over bandwidth-constrained mobile connections. The standard supports scalable coding configurations where base layers provide basic 3D quality while enhancement layers add additional views or higher resolution. This scalability is particularly important for adaptive streaming scenarios where network conditions may vary. The technology integrates with 3GPP's multimedia delivery frameworks including Dynamic Adaptive Streaming over HTTP (DASH) and Multimedia Broadcast/Multicast Service (MBMS) for both unicast and broadcast distribution of 3D content.

Purpose & Motivation

3D-HEVC was developed to address the growing demand for immersive 3D video services while overcoming the bandwidth limitations of mobile networks. Traditional 2D video coding approaches proved inefficient for 3D content because they treated multiple views as independent video streams, failing to exploit the high correlation between different perspectives of the same scene. This resulted in approximately double the bandwidth requirement for stereo video compared to 2D video, making 3D services impractical for mass deployment over cellular networks.

The technology emerged during a period of increasing interest in 3D television and immersive media experiences. Previous approaches to 3D video delivery either used frame-compatible formats that sacrificed resolution or transmitted full left and right views independently with inefficient compression. 3D-HEVC solved these problems by introducing specialized coding tools that recognize the geometric relationships between views, enabling much higher compression efficiency. This made it possible to deliver high-quality 3D video at bitrates only slightly higher than equivalent 2D content.

Another key motivation was the need to support advanced 3D display technologies including autostereoscopic displays that require multiple views for glasses-free 3D viewing. These displays typically need 5-9 different views to provide proper 3D perception from various viewing angles. Transmitting all these views independently would be prohibitively expensive in terms of bandwidth. 3D-HEVC's ability to efficiently code multiple views and synthesize intermediate views at the receiver enabled practical deployment of such advanced 3D display systems over mobile networks.

Key Features

• Disparity-compensated prediction for inter-view coding
• Depth map coding with specialized depth modeling modes
• Support for multi-view video plus depth (MVD) format
• View synthesis optimization for intermediate view generation
• Scalable coding with base and enhancement layers
• Integration with 3GPP multimedia delivery frameworks

Evolution Across Releases

Defining Specifications