3D Image Encoding via Parallax Data Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current techniques for encoding and decoding three-dimensional images face challenges with high computational load and data volume, even when limited to a minute viewpoint movement range, due to the need for geometry calculations and extensive data encoding.
Innovation Solution
The method employs a camera setup with a main camera and sub-cameras to encode three-dimensional image information into multiple two-dimensional images through 'shift + pixel value comparison', allowing for reduced calculation and data storage by generating parallax data without geometry calculations, and using transparent regions to optimize data storage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If geometry calculation is performed on encoded three-dimensional image information to achieve viewpoint movement, then the decoding can be accomplished, but the computational load becomes enormous
Solution Approach 1:
The patent extracts only the necessary parallax information from multiple viewpoint images, storing it in a compact format. Instead of performing complete geometry calculations on all three-dimensional image data, the system extracts and stores only the differential parallax data needed for viewpoint movement, significantly reducing the computational load during decoding while maintaining the capability to generate views from different angles.
2Adaptability or versatility
If all three-dimensional image information is encoded to enable comprehensive viewpoint movement, then the viewpoint flexibility is improved, but the data volume becomes enormous
Solution Approach 1:
The patent uses a main viewpoint image as the base and creates simplified copies or representations of the three-dimensional scene by encoding only the parallax differences relative to this main view. Instead of storing complete image data from multiple viewpoints, the system stores one full-resolution main image plus compact parallax difference data, which can be used to synthesize other viewpoint images on demand.
3Reliability
If conventional encoding methods are used for three-dimensional images, then complete image information is preserved, but the encoding complexity and data requirements increase significantly
Solution Approach 1:
The patent segments the three-dimensional image encoding into two distinct parts: a main viewpoint image that contains the complete base scene information, and separate parallax data that contains only the depth and viewpoint variation information. This segmentation allows the main image to be encoded using conventional efficient image compression methods, while the parallax data is stored in a compact format that enables viewpoint synthesis without requiring complex encoding of the entire three-dimensional scene.
Data Source
Figure 1
Figure 2~3
Figure 4(a)~4(b)
AI summary
An image encoding method for an encoding system includes receiving a first image captured by a main camera and a second image captured by a sub camera provided in the same plane; comparing pixel values of pixels at first coordinates of each of the first and second images to obtain a first difference value; determining the first coordinates are in surface or back region in case of the first difference value is not more than or more than a first threshold value; pixel-shifting the second image by a first shift amount; and comparing the pixel value at the first coordinates of the first image with that of the second image after the pixel shift to obtain a second difference value, determining whether the first coordinates are in the surface or back region based on the second difference value, and recording into a second frame different from the first frame.