Multi-Viewpoint Image Encoding Using Ordered Reference Lists
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In 3D coding systems, the coding efficiency of images from multiple viewpoints is not adequately improved due to the arbitrary assignment of view IDs, which makes it difficult to recognize camera arrangements and utilize position information effectively.
Innovation Solution
An encoding device and method that sets first and second image specifying information for viewpoints in specific orders in reference lists, allowing for improved coding based on these ordered lists, and a decoding device and method that decodes data using these ordered lists to enhance coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If view IDs are assigned arbitrarily to camera viewpoints, then flexibility in viewpoint assignment is improved, but the ability to recognize camera arrangement and improve coding efficiency deteriorates
Solution Approach 1:
The patent changes the parameter of view ID assignment from arbitrary to sequential based on camera position. By establishing a rule that view IDs are assigned sequentially from one end of the camera array, the system transforms the flexible but inefficient arbitrary assignment into a structured assignment that enables efficient coding while maintaining adaptability to different camera configurations
Solution Approach 2:
The patent introduces a new dimension of spatial ordering by arranging viewpoints in a sequential order that reflects their physical camera positions. This dimensional organization of view IDs allows the coding system to exploit spatial relationships between adjacent cameras, improving prediction accuracy and coding efficiency
2Ease of operation
If reference images are registered in the same order in both L0 and L1 lists, then list management simplicity is improved, but coding efficiency improvement deteriorates
Solution Approach 1:
The patent segments the reference image lists into two distinct ordering schemes: L0 list orders reference images in increasing view ID sequence while L1 list orders them in decreasing view ID sequence. This segmentation allows each list to be optimized for its specific prediction direction, improving coding efficiency while maintaining clear management rules
Solution Approach 2:
The patent applies inversion by registering reference images in opposite orders in L0 and L1 lists. While L0 uses increasing view ID order, L1 uses decreasing view ID order, allowing the system to efficiently access reference images from both temporal directions with optimized ordering for each prediction mode
Data Source
AI summary
The present technology relates to an encoding device, an encoding method, a decoding device, and a decoding method capable of improving the coding efficiency of an image of a plurality of viewpoints. A motion parallax prediction/compensation unit sets first image specifying information specifying a first image of a viewpoint present in a first direction from a viewpoint of a current coding image and second image specifying information specifying a second image of a viewpoint present in a second direction opposite to the first direction in a list (L0) in first order and sets the first image specifying information and the second image specifying information in a list (L1) in second order that is different from the first order. The motion parallax prediction/compensation unit and a calculation unit generate coded data by coding the current coding image based on the lists (L0 and L1). The present technology, for example, can be applied to an encoding device that codes a 3D image and the like.