Depth Lookup Table Prediction for 3D Video Coding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for coding depth lookup tables (DLTs) in 3D video processing are inflexible and do not fully utilize signal characteristics, leading to suboptimal coding efficiency, particularly in terms of temporal and spatial updating, resulting in reduced depth map quality in dependent views.
Innovation Solution
The proposed solution involves using DLT prediction, where only the difference between the current and a reference DLT is signaled, exploiting temporal and inter-view redundancies to reduce bitrate, and employing methods such as XOR operations for efficient coding and decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If DLT is signaled only once per sequence in SPS, then the overhead for DLT signaling is kept low, but the flexibility for temporal and spatial updating is reduced
Solution Approach 1:
The patent segments the DLT signaling into multiple levels: sequence-level (SPS) for common depth values, picture-level (PPS) for view-specific updates, and slice-level for fine-grained updates. This segmentation allows the system to maintain low overhead for common values while providing flexibility for updates when needed, resolving the contradiction between minimal overhead and updating flexibility.
Solution Approach 2:
The patent implements dynamic DLT updating mechanisms where the DLT can be updated at different levels (SPS, PPS, slice) depending on the coding scenario and histogram changes. This dynamic approach allows the system to adapt between static (low overhead) and dynamic (high flexibility) modes, resolving the contradiction between fixed overhead and flexible updating.
2Reliability
If DLT values are updated regularly in temporal direction, then histogram changes over time are captured, but the coding complexity and overhead increase
Solution Approach 1:
The patent applies local quality updates by allowing DLT updates at specific locations in the coding structure (PPS level for view-specific updates, slice level for region-specific updates) rather than requiring uniform updates across the entire DLT. This enables targeted updates that improve depth map quality for specific views or regions while avoiding the complexity of updating the entire DLT regularly.
Solution Approach 2:
The patent changes the parameter of update frequency and scope based on coding conditions. Instead of fixed regular updates, the system adjusts update parameters (when and where to update) based on histogram characteristics and coding scenarios, reducing unnecessary updates and associated complexity while maintaining quality when needed.
3Quantity of substance
If all dependent views inherit the base view's DLT, then the overhead is minimized, but the depth map quality in dependent views with different histograms deteriorates
Solution Approach 1:
The patent segments the inheritance mechanism into selective inheritance models: base view inheritance for common depth values, view-specific updates for view-dependent variations, and slice-level updates for region-specific adjustments. This segmentation allows the system to minimize overhead for common values while providing precision updates for views with different histograms, resolving the contradiction between low overhead and high accuracy.
Solution Approach 2:
The patent applies local quality by allowing different levels of DLT inheritance and updating for different views and regions. View-specific and slice-level updates enable the system to adapt the DLT quality locally for views with different histogram characteristics, maintaining depth map reconstruction accuracy without requiring full DLT retransmission for all views.
Data Source
AI summary
The invention relates to a method (100) for coding a depth lookup table, DLT, (201, 211), the depth lookup table comprising depth value information of at least a part of a 3D picture, the method (100) comprising: selecting (101) a reference depth lookup table (203, 213); determining (103) a difference depth lookup table (205, 215) based on a comparison between depth value information of the depth lookup table to be coded (201, 211) and depth value information of the reference depth lookup table (203, 213); and coding (105) depth value information of the difference depth lookup table (205, 215) according to a predetermined coding rule.