Integral Image Residual Coding for 3D Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for compressing integral images in integral imaging systems face challenges due to high resolution requirements, leading to poor performance and increased data volume, particularly in coding and decoding processes for 3D video applications.
Innovation Solution
A method that decomposes integral images into views, codes each view, decodes them, and recomposes the image with residual data coding, optimizing data quantity and quality by selecting optimal views and quantization parameters, and applying image transformations to reduce data loss.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the resolution of the integral image is increased and the number of micro-images is increased to achieve high resolution and large number of points of view, then the quality of 3D visualization is improved, but the size of the integral image increases much too large
Solution Approach 1:
The integral image is divided into multiple micro-images, each representing a different perspective. By segmenting the high-resolution integral image into smaller micro-images, the system can manage and compress each segment separately, reducing the overall computational burden and data volume while maintaining the ability to reconstruct high-resolution views.
Solution Approach 2:
The patent transforms the problem from compressing a single large 2D integral image to compressing multiple smaller 2D micro-images that can be arranged in a 3D structure. This dimensional transformation allows for more efficient compression by exploiting the spatial relationships between micro-images and applying compression algorithms to each smaller segment independently.
2Measurement precision
If the size of each micro-image and the number of micro-images are increased to achieve high resolution and large number of points of views, then the quality of 3D visualization is improved, but the compression performance becomes poor
Solution Approach 1:
By segmenting the integral image into multiple micro-images, the system can apply compression algorithms to each smaller segment independently. This segmentation reduces the computational complexity of compression while maintaining the ability to reconstruct high-resolution views, thereby improving compression performance.
Solution Approach 2:
The patent merges multiple compressed micro-images to reconstruct the full integral image. By combining the compressed segments at the decoder side, the system achieves both high compression ratios and high reconstruction quality, resolving the contradiction between compression performance and image quality.
3Device complexity
If conventional 2D coding methods are applied directly to the integral image, then the coding process is simple, but the compression performance is poor due to high volume of information
Solution Approach 1:
The patent segments the integral image into multiple micro-images before applying 2D compression algorithms. This segmentation reduces the data volume that needs to be processed by each compression algorithm instance, improving compression performance while keeping the overall process relatively simple.
Solution Approach 2:
The approach transforms the problem from applying 2D compression to a single large image to applying 2D compression to multiple smaller images arranged in a 3D structure. This dimensional change enables better compression performance by exploiting the spatial relationships between micro-images while maintaining the simplicity of 2D coding methods.
4Productivity
If MVC technique is used to compress the series of views, then the compression efficiency is improved, but the number of views is limited and high resolution is required for each view
Solution Approach 1:
The patent segments the integral image into multiple micro-images that can be treated as views. By segmenting the original image into smaller units, the system can apply MVC techniques to these segments, achieving better compression efficiency while reducing the resolution requirement for each individual view compared to the original full-resolution image.
Data Source
AI summary
The invention concerns the encoding of at least one current integral image (IIj) captured by an image capture device, comprising the steps consisting of: —decomposing (C1) the current integral image into at least one frame (Vu) representing a given perspective of a scene and, from at least one image capturing parameter associated with the image capture device, —encoding (C2) said at least one frame, —decoding (C4) said at least one frame, —recomposing (C5) the current integral image from said at least one decoded frame by applying an inverse decomposition of said decomposition of the integral image and from said at least one image capturing parameter associated with the image capture device, said encoding method being characterised in that it implements the steps consisting of: —determining (C6) a residual integral image by comparing said at least one current integral image with said recomposed integral image, —encoding (C7) the data associated with the residual integral image and said at least one image capturing parameter associated with the image capture device.