Quadtree Disocclusion Atlas Packing for Multi-View Video Synthesis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-view video systems face issues with disocclusions and gaps in synthesized views due to reduced view numbers, leading to noticeable visual artifacts, which are not effectively addressed by current methods.
Innovation Solution
The use of a disocclusion atlas to encode and transmit only texture and depth information for occluded image fragments, minimizing redundant data and enhancing compression efficiency by stabilizing consecutive atlas layouts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If the total number of views to be encoded or transmitted is reduced or minimized, then transmission bandwidth and storage requirements are decreased, but regions of holes in synthesized views become relatively large and numerous, creating noticeable visual artifacts
Solution Approach 1:
The patent segments the synthesized view into multiple spatial regions, identifying and categorizing different types of holes (disocclusion holes, warping holes, etc.) separately. This segmentation allows targeted processing of different hole types using appropriate filling strategies, improving overall artifact reduction effectiveness while maintaining efficient transmission with minimal views.
Solution Approach 2:
The patent performs preliminary identification and classification of hole regions during the view synthesis process, before final rendering. By pre-marking disocclusion regions and other hole areas in the synthesized view, the system prepares the data structure for efficient artifact correction, allowing bandwidth-limited transmission to still produce high-quality views with minimal visual artifacts.
2Object-affected harmful factors
If conventional hole filling methods are used to fill disocclusion regions, then gaps in synthesized views are filled, but processing complexity and computational load increase significantly
Solution Approach 1:
The patent applies different processing strategies to different spatial regions and hole types based on their local characteristics. Disocclusion holes are handled differently from warping holes, and boundary regions are treated differently from interior regions. This localized approach fills gaps effectively while avoiding uniform complex processing across the entire image, thereby reducing overall computational load.
Solution Approach 2:
The patent changes key parameters such as depth values and texture coordinates in disocclusion regions to enable efficient hole filling. By modifying depth maps to mark disoccluded areas and adjusting texture sampling parameters, the system achieves effective gap filling through parameter adjustments rather than computationally intensive pixel-level processing, reducing device complexity.
3Manufacturing precision
If depth image discontinuities are corrected to improve view synthesis quality, then gaps caused by depth errors are reduced, but processing time and computational resources increase
Solution Approach 1:
The patent applies partial depth correction only to regions where depth discontinuities are detected and where they actually cause visible gaps in the synthesized view. Rather than processing the entire depth map uniformly, the system selectively corrects problematic regions, achieving sufficient depth image quality improvement while minimizing processing time and computational resource consumption.
Data Source
AI summary
Occluded image fragments are sorted in size. The largest image fragment is used to size a quadtree node in a layout mask for a disocclusion atlas used to store the image fragments. The sorted image fragments are stored into the disocclusion atlas using the layout mask such as each image fragment is hosted with a best fit quadtree node in the disocclusion atlas. A video signal may be generated by encoding one or more reference images and the disocclusion atlas storing the image fragments. The image fragments can be used by a recipient device to fill disoccluded image data in disoccluded spatial regions in a display image synthesized from the reference images.