3D Video Stream Camera Parameter Signaling via ViewOldxList
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current 3D-HEVC designs face issues with camera parameter signaling, particularly in handling auxiliary pictures and bitstream extraction, where scale and offset values are not efficiently managed, leading to parsing dependencies and unnecessary rewriting of slice headers.
Innovation Solution
Direct indication of relative camera parameters associated with coded views, using a table to list reference views, allowing for efficient signaling and decoding without relying on indirect indexing, which reduces computational overhead and maintains robustness against changes in view order.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If indirect indexing through ViewOldxList is used to signal camera parameters, then the system maintains compatibility with auxiliary picture layers, but it creates parsing dependencies and requires extensive rewriting of slice headers when layers are discarded
Solution Approach 1:
The patent segments the camera parameter signaling into two independent parts: a static ViewOldxList in the VPS that defines view mappings, and dynamic scale/offset pairs in slice headers that reference these views. This segmentation allows independent handling of each component, eliminating the parsing dependency where slice header interpretation relied on the complete iteration through ViewOldxList.
Solution Approach 2:
The ViewOldxList is preliminarily established in the VPS before slice header processing. Each entry in ViewOldxList pre-defines the mapping between layer IDs and view order indices, allowing slice headers to directly reference views using these pre-established mappings without requiring runtime iteration or dependency on auxiliary layer configurations.
2Reliability
If scale and offset values are signaled for all views in ViewOldxList, then completeness of camera parameters is ensured, but bandwidth is wasted signaling parameters for views that are not actually used
Solution Approach 1:
The patent applies local quality by making camera parameter signaling view-specific and demand-driven. Instead of uniformly signaling parameters for all views in ViewOldxList, the system signals scale and offset pairs only for the specific number of views (num_views_minus_one) that are actually needed for depth-to-disparity conversion in each slice, optimizing bandwidth usage according to local requirements.
Solution Approach 2:
The system dynamically changes the number of signaled camera parameter pairs based on actual needs. The num_views_minus_one parameter in slice headers allows flexible adjustment of how many scale/offset pairs are transmitted, enabling the system to adapt parameter signaling quantity to the specific decoding requirements of each slice while maintaining completeness for needed views.
3Adaptability or versatility
If camera parameters are signaled in slice headers for each view, then flexibility in handling different view configurations is achieved, but the number of parameters to parse and manage increases significantly
Solution Approach 1:
The ViewOldxList acts as an intermediary between layer IDs and view order indices. Instead of directly managing complex mappings between all possible layer combinations and views, the system uses ViewOldxList as a lookup table that simplifies the relationship, allowing slice headers to reference views through a straightforward indexing mechanism rather than requiring complex parsing logic.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A 3D video data stream 2 comprising a set of coded views 4, 4a, 4b coded in the 3D video data stream in temporal units is shown. The 3D video data stream further comprises a parameter set 6 comprising a first table 10 having an entry for each coded view 4, 4a, 4b, which indicates by indexing for the respective coded view a number of reference views for which a relative camera parameter is present in the data stream, wherein a scope of the parameter set (6) is beyond the scope of the temporal units.