MultiView High Efficiency Video Coding

Description

MV-HEVC, or Multi-view High Efficiency Video Coding, is a standardized video coding profile within the HEVC (H.265) family, specifically designed for compressing video sequences captured from multiple cameras viewing the same scene from slightly different perspectives. The core architectural principle is based on exploiting the high degree of inter-view redundancy—similarities between different camera views—in addition to the temporal redundancy within a single view sequence. This is achieved through advanced prediction techniques.

The codec works by designating one view as the base view, which is encoded using a standard HEVC single-layer codec, ensuring backward compatibility with legacy HEVC decoders. Additional dependent views are then encoded by heavily referencing the base view and other already encoded dependent views. Key coding tools include inter-view motion prediction, where motion vectors can be predicted from other views; inter-view residual prediction; and disparity-compensated prediction, which is analogous to motion compensation but for spatial differences between views. The bitstream structure supports the encapsulation of multiple views, along with necessary parameter sets that describe the relationship between them (e.g., camera parameters, view order).

In a 3GPP service context, such as MBMS (Multimedia Broadcast Multicast Service) or streaming, MV-HEVC enables the efficient delivery of 3D or multi-view content. The network transmits a single, integrated bitstream. A compatible client device can decode the base view for 2D playback or decode multiple views to render stereoscopic 3D video for compatible displays or to enable interactive viewpoint switching. Key components are the MV-HEVC decoder, which implements the inter-view prediction tools, and the signaling within the media presentation description (e.g., DASH MPD) to indicate the availability of multi-view representations. This allows adaptive streaming clients to select the appropriate number of views based on device capability, network conditions, and user preference.

Purpose & Motivation

MV-HEVC was developed to meet the demands of emerging immersive video applications, such as 3DTV, free-viewpoint video, and virtual reality, within the constraints of mobile network bandwidth. Before its standardization, transmitting multiple independent video streams for each camera view was prohibitively bandwidth-intensive, limiting the feasibility of such services over wireless links.

The motivation stemmed from the success of earlier multi-view coding in H.264/AVC (MVC) and the need for greater efficiency offered by the newer HEVC codec. HEVC already provided roughly 50% bitrate savings over H.264 for single-view video. MV-HEVC extends these savings to the multi-view domain by applying HEVC's advanced coding tools across views. This directly addresses the limitation of the previous MVC standard by providing a more efficient, modern codec foundation, enabling high-quality multi-view services on bandwidth-limited 4G and 5G networks. It was a key enabler for 3GPP's work on advanced media services in releases focusing on enhanced mobile broadband and media delivery.

Key Features

• Efficient compression of multiple video views by exploiting inter-view redundancy
• Backward compatibility through a base view decodable by standard HEVC decoders
• Advanced prediction tools including inter-view motion prediction and disparity-compensated prediction
• Support for stereo 3D video and interactive free-viewpoint playback
• Integration with 3GPP media delivery frameworks like DASH and MBMS
• Scalable view selection allowing clients to decode a subset of views based on capability

Evolution Across Releases

Defining Specifications