Inter-view coding mode indicator for multiview video
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multiview Video Coding (MVC) technologies face inefficiencies in encoding and decoding processes due to the complexity of inter-view prediction methods, particularly in scenarios where camera positions differ significantly, leading to increased bitstream size and computational requirements.
Innovation Solution
The implementation of inter-view coding modes in MVC systems, where two-dimensional motion vectors are reduced to one-dimensional vectors based on camera alignment, allowing for efficient encoding and decoding by indicating the inter-view coding mode through mode indicators in the bitstream, thereby reducing the number of bits required for motion vector components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If two-dimensional motion vectors are used for inter-view prediction, then prediction accuracy is improved, but bitstream size and computational complexity increase
Solution Approach 1:
The patent segments the motion vector into two parts: a predicted component derived from the base view and a residual component that needs to be encoded. This segmentation allows the encoder to exploit the correlation between views while reducing the number of bits needed to represent the full motion vector, thus resolving the contradiction between prediction accuracy and bitstream size.
Solution Approach 2:
The patent transitions from encoding full two-dimensional motion vectors to encoding only the residual difference between predicted and actual motion vectors. This dimensional reduction in the encoded data (from 2D to a smaller residual space) maintains prediction accuracy while reducing bitstream size and computational complexity.
2Measurement precision
If full two-dimensional motion vectors are encoded, then motion estimation precision is improved, but the number of bits required increases
Solution Approach 1:
The patent extracts and encodes only the essential residual information - the difference between the predicted motion vector from the base view and the actual motion vector. By taking out only the necessary residual components rather than encoding the full motion vector, the patent reduces the number of bits required while maintaining motion estimation precision.
Solution Approach 2:
The patent applies partial encoding by only encoding the residual portion of the motion vector rather than the complete vector. This partial action is sufficient to maintain precision because the base view prediction provides the majority of the motion information, and only the residual needs to be transmitted.
3Productivity
If inter-view prediction uses reference pictures from other views, then coding efficiency is improved, but complexity of the decoding process increases
Solution Approach 1:
The patent performs preliminary prediction by deriving motion vectors from the base view before encoding the residual. This preliminary action simplifies the decoding process because the decoder can reconstruct the predicted motion vector from the base view and then add the transmitted residual, rather than performing complex inter-view prediction from scratch.
Data Source
AI summary
Systems, devices and methods are described including determining an inter-view coding mode for at least a portion of an image frame, specifying a corresponding value of an inter-view coding mode indicator, and providing the mode indicator in a bitstream that includes an encoded motion vector associated with the image frame portion. A first value of the mode indicator corresponds to a first inter-view coding mode where the encoded motion vector includes components in multiple dimensions. A second value of the mode indicator corresponds to a second inter-view coding mode where the encoded motion vector components include components in only one dimension.