Depth Map Enrichment for Occlusion Precision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The finite resolution of depth maps leads to approximations in determining occlusion and shadowing, resulting in artifacts like bias aliasing and peter-panning, especially when fragments belong to the same surface.
Innovation Solution
Enhancing the depth map content by associating each element with additional information representing the variation of depth within the element, allowing for more precise depth retrieval at any point within the element.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a depth map with finite resolution is used to determine occlusion and shadowing, then the processing is simplified and faster, but the precision of depth determination deteriorates, leading to aliasing artifacts
Solution Approach 1:
The patent divides each depth map element into multiple sub-elements or samples, allowing independent depth determination for different regions within the same element. This segmentation enables more precise depth measurement without requiring the entire depth map to be processed at higher resolution, thus maintaining processing speed while improving precision.
Solution Approach 2:
The patent introduces an additional dimension of sampling within each depth map element, transitioning from a single depth value per element to multiple depth values at different positions or with different weights. This dimensional expansion allows precise depth determination while keeping the overall depth map structure intact for efficient processing.
2Measurement precision
If a bias value is introduced to overcome bias aliasing, then the accuracy of occlusion determination improves, but a new artifact known as peter-panning artifact is introduced
Solution Approach 1:
The patent applies different depth sampling strategies to different regions of the depth map based on local characteristics. Instead of using a uniform bias value across the entire depth map, the system adapts the depth determination method to local areas, applying bias correction only where necessary and using direct comparison where possible, thereby avoiding the peter-panning artifact while maintaining accuracy.
Solution Approach 2:
The patent dynamically adjusts depth determination parameters based on local scene characteristics and depth map properties. By changing the determination method from fixed bias addition to adaptive sampling and comparison, the system eliminates the peter-panning artifact while maintaining accurate occlusion detection.
3Measurement precision
If depth information is sampled at multiple positions within an element, then the precision of depth retrieval improves, but the data storage and processing complexity increases
Solution Approach 1:
The patent samples depth information at multiple positions within each element, but only to the extent necessary to achieve the desired precision. Instead of uniformly sampling the entire element, the system applies selective sampling based on the specific requirements of the rendering task, storing only the necessary depth information and using it efficiently during processing.
Solution Approach 2:
The patent pre-computes and stores depth information at multiple sampled positions within each element during the depth map generation phase. This preliminary action allows the system to retrieve precise depth values during rendering without performing complex calculations in real-time, reducing processing complexity while maintaining high precision.
Data Source
Figure 1
Figure 2~3B
Figure 4~5
AI summary
A method for enriching the content associated with a first element (21) of a depth map (10) the depth map (10) being associated with a scene (1) according to a point of view. Thereafter, at least a first information representative of a variation of depth in the first element (21) in the space of the depth map is established.