Image Processing Subband Division for RAW Data Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems fail to efficiently compress RAW data from image pickup elements with alternately shifted pixel positions, leading to decreased compression efficiency and increased noise due to separation of correlated images and lack of hardware sharing across different pixel arrangements.
Innovation Solution
The system performs subband division of image data with pixels from adjacent lines or columns as units, using wavelet or Haar transforms to maintain shifted pixel positions during compression, allowing for efficient coding across various pixel arrangements including Bayer, double density Bayer, and oblique arrangements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If green pixels are separated into G1 and G2 components by sub-sampling, then compression can be performed on each component separately, but the correlation between separated images cannot be used and compression efficiency is decreased
Solution Approach 1:
The patent segments the image data into multiple components (G1, G2, R, B) based on pixel positions, but processes them as a unified structure rather than completely separate images. This allows correlation between adjacent pixels to be preserved while enabling component-specific processing.
Solution Approach 2:
The patent merges the processing of G1 and G2 components by applying wavelet transform to the combined image data structure, maintaining the spatial correlation between green pixels from different color filter positions while achieving efficient compression.
2Productivity
If wavelet transform is applied to the entire picture, then very high compression efficiency can be achieved, but the existing system separates the entire picture into separate pictures and does not exert the inherently high compression efficiency
Solution Approach 1:
The patent creates a universal processing structure that handles multiple pixel arrangements (Bayer, double density Bayer, oblique arrangements) using the same wavelet transform algorithm, making the compression system adaptable to different sensor configurations without requiring separate processing paths.
Solution Approach 2:
The patent changes the processing parameter from separate image processing to unified image data processing, applying wavelet transform to the entire picture while preserving the Bayer pattern structure, thereby achieving high compression efficiency.
3Device complexity
If simple pixel discrete reduction is performed for half-resolution display, then processing is simplified, but the image displayed on the viewfinder is affected by aliasing noise
Solution Approach 1:
The patent applies wavelet transform as a preliminary action before resolution reduction, which decomposes the image into frequency subbands. This allows for proper downsampling without aliasing by filtering out high-frequency components that would cause noise in the reduced-resolution display.
4Manufacturing precision
If different compression systems are used for different pixel arrangements, then each arrangement can be optimized, but hardware cannot be shared
Solution Approach 1:
The patent designs a universal compression system based on wavelet transform that can handle multiple pixel arrangements (Bayer, double density Bayer, oblique arrangements) through a single hardware architecture, enabling hardware sharing while maintaining optimization for each arrangement type.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances compression efficiency, reduces noise, and enables common hardware processing for different pixel arrangements, achieving high-resolution image handling with reduced hardware requirements and improved image quality.
Implementation Method 1
A wavelet transform in particular can achieve a very high compression efficiency by subband division of an entire picture.
Implementation Method 2
The system performs subband division of image data with pixels from adjacent lines or columns as units, using wavelet or Haar transforms to maintain shifted pixel positions during compression
Data Source
AI summary
Disclosed herein is an image processing device including a subband dividing section configured to perform subband division of image data of a color whose pixel positions are alternately shifted from each other, the image data being included in image data output from an image pickup element of a pixel arrangement in which the pixel positions of at least one color of three primary colors are alternately shifted from each other in one of a horizontal direction and a vertical direction, with pixels of two upper and lower lines adjacent to each other or pixels of two left and right columns adjacent to each other as a unit.