Depth-to-Disparity Vector Conversion for 3D Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current multi-view video coding techniques face inefficiencies in storage and transmission due to high bandwidth requirements, and existing disparity vector derivation methods are complex and resource-intensive, particularly in depth-to-disparity conversions.
Innovation Solution
A method that simplifies the depth-to-disparity vector conversion by using a single converted disparity vector for a conversion region, derived from maximum depth values, to unify and reduce processing overhead across various coding tools and modes, such as Inter, Skip, and Direct modes, while maintaining performance by reducing memory accesses and bit rate.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional depth-to-disparity conversion is performed for each motion prediction sub-block independently, then the coding precision is improved, but the device complexity and processing overhead increase
Solution Approach 1:
The patent merges the depth-to-disparity conversion process by performing it once at the macroblock level and reusing the resulting disparity vector for all motion prediction sub-blocks within that macroblock. This combining approach reduces the number of conversions from multiple per-macroblock to a single conversion, thereby reducing device complexity and processing overhead while maintaining acceptable coding precision through the shared disparity vector.
Solution Approach 2:
The patent creates a universal disparity vector at the macroblock level that serves multiple functions across different motion prediction sub-blocks. This single disparity vector is reused universally for all sub-blocks within the macroblock, eliminating the need for separate conversion processes and reducing overall system complexity while still providing effective disparity compensation for each sub-block.
2Adaptability or versatility
If multiple cameras are used to capture multiple video sequences, then the viewing experience is improved, but the storage space and transmission bandwidth requirements increase
Solution Approach 1:
The patent uses disparity vectors derived from depth maps to create virtual copies of video content from different viewpoints. Instead of storing and transmitting multiple actual camera sequences, the system generates synthetic views by displacing pixels based on disparity information, effectively creating compressed virtual copies that reduce storage and bandwidth requirements while maintaining multi-view functionality.
Solution Approach 2:
The patent introduces depth maps as an intermediary representation that captures the essential 3D structure of the scene. This intermediary depth information serves as a compact representation from which multiple view sequences can be synthesized, reducing the need to store and transmit full multi-view video sequences while still enabling rich viewing experiences through view synthesis prediction.
3Measurement precision
If depth-to-disparity conversion is performed for each sub-block, then the measurement precision is improved, but the loss of time in processing increases
Solution Approach 1:
The patent performs the computationally intensive depth-to-disparity conversion action in advance at the macroblock level, before the actual motion prediction and compensation steps for each sub-block. This preliminary conversion produces a disparity vector that is then reused across all sub-blocks, eliminating the need for repeated conversions and significantly reducing processing time while maintaining sufficient disparity accuracy for all sub-blocks within the macroblock.
Data Source
AI summary
A method and apparatus using a single converted DV (disparity vector) from the depth data for a conversion region are disclosed. Embodiments according to the present invention receive input data and depth data associated with a conversion region of a current picture in a current dependent view. The conversion region is checked to determine whether it is partitioned into multiple motion prediction sub-blocks. If the conversion region is partitioned into multiple motion prediction sub-blocks, then a single converted DV from the depth data associated with the conversion region is determined and each of the multiple motion prediction sub-blocks of the conversion region is processed according to a first coding tool using the single converted DV. If the conversion region is not partitioned into multiple motion prediction sub-blocks, the conversion region is processed according to the first coding tool or a second coding tool using the single converted DV.