Depth Map Value Mapping for 3D Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional 3D video coding methods fail to effectively handle non-coplanar camera arrangements in 3D video coding, leading to inaccurate view synthesis and reduced coding performance due to imperfect depth map estimation and occlusions in inter-view prediction.
Innovation Solution
A configurable depth map value mapping technique is used to preprocess depth maps, enhancing homography mappings and improving prediction precision without increasing coding costs, by focusing on the range where most depth map values are located and applying a spatial filter to adapt and filter depth information values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If depth maps are estimated from texture data or pre-processed using conventional methods, then the coding process can proceed, but the depth map quality is insufficient leading to inaccurate view synthesis and reduced coding performance
Solution Approach 1:
The patent applies a configurable depth map value mapping that transforms depth values from the original range to an adapted range with higher resolution. This parameter transformation increases the precision of depth representation, allowing for more accurate homography mappings and view synthesis without requiring changes to the underlying estimation algorithms.
Solution Approach 2:
The depth map value mapping is performed as a preprocessing step before inter-view prediction and view synthesis. By adapting the depth values in advance, the patent ensures that subsequent operations work with high-precision depth information, improving overall view synthesis accuracy without adding complexity to the main coding pipeline.
2Loss of information
If the full range of depth quantization levels is used for inter-view prediction, then all depth information is preserved, but the precision is insufficient in the range where most depth values are located
Solution Approach 1:
The patent applies different mapping strategies to different ranges of depth values. The configurable depth map value mapping focuses on increasing precision in the reduced range where most depth values are concentrated, while still maintaining coverage of the full original range. This local quality enhancement ensures high precision where it matters most without losing information in other ranges.
Solution Approach 2:
The patent effectively adds a dimensional transformation by mapping depth values from one range space to another. The depth value adaptation function transforms the distribution of depth values, spreading them across an adapted range that provides higher resolution in the region of interest, thereby improving precision without reducing the informational content.
3Ease of manufacture
If conventional block-based view synthesis prediction is used for coplanar camera arrangements, then the method is simple to implement, but it fails for non-coplanar camera arrangements where pixels shift differently between views
Solution Approach 1:
The patent uses configurable depth map value mapping to enable precise per-pixel depth representation, which is essential for handling non-coplanar camera arrangements. By adapting depth values to a higher resolution scale, the system can accurately model different pixel shifts between views while maintaining a relatively simple implementation based on existing block-based prediction frameworks.
Solution Approach 2:
The patent introduces dynamic adaptability through the configurable mapping parameters that can be adjusted based on the specific camera arrangement. The depth value adaptation function can be tuned to match the geometric relationships in non-coplanar setups, allowing the system to adapt to different camera configurations without requiring completely different algorithms.
4Productivity
If a reference depth map is used for generating dependent texture or depth map, then the process is efficient, but inaccurate depth maps lead to poor prediction regions selection and reduced coding efficiency
Solution Approach 1:
The patent transforms the depth value parameters to a higher resolution scale before using them for prediction. This parameter change allows for more precise calculation of prediction regions and better selection of matching areas, thereby improving prediction accuracy while maintaining the efficiency of reference-based generation methods.
Solution Approach 2:
The adapted depth map values serve as an intermediary representation that bridges the gap between the original low-precision depth map and the high-precision requirements of view synthesis. By using the mapped depth values as an intermediate step, the system achieves better prediction accuracy without fundamentally changing the reference-based generation approach.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to video coding with inter-view prediction. A reconstructed depth map of a reference view is obtained, quantized according to a predetermined reference quantization. A reduced range information is obtained, wherein the reduced range information defines a reduced range of depth quantization levels available for the inter-view prediction within the range of quantization levels defined by the predetermined reference quantization. The distance information corresponding to the reduced range of depth quantization levels is enhanced by an adaptation function, which is defined based on the reduced range. The enhanced distance information is used for inter-view prediction, leading to improved coding efficiency.