Image Color-Difference Encoding with Co and Cg Statistics for Lower Bit Usage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image encoding methods face challenges in removing redundant information between color signals, which affects encoding and compression performance and increases computation complexity, particularly in applications with high real-time requirements like self-driving.
Innovation Solution
A method involving determining color difference signals and average values for adjacent pixels in a 2*2 image block to reduce redundant information, thereby optimizing bit usage and computation complexity, including steps to calculate and output difference values, average values, and color components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If traditional color conversion methods are used to encode color signals, then encoding performance is maintained, but transmission bandwidth requirement increases and computation complexity increases
Solution Approach 1:
The patent extracts and removes redundant information from color signals by calculating color difference signals (Cb, Cr) and their statistical properties (mean Co, variance Cg). Only the essential difference value Cg and average value Co are transmitted instead of full color difference signals, thereby reducing transmission bandwidth while preserving encoding performance.
Solution Approach 2:
The patent transforms the color signal representation from transmitting full color difference signals (Cb, Cr) to transmitting statistical parameters (mean Co, variance Cg). This parameter transformation reduces the amount of data required for transmission while maintaining the ability to reconstruct color information at the decoder.
2Loss of information
If traditional color conversion methods are used to encode color signals, then encoding performance is maintained, but computation complexity increases
Solution Approach 1:
The patent extracts only the essential statistical parameters (mean and variance) from color difference signals, removing redundant computational operations. By calculating and transmitting only Co and Cg instead of full Cb and Cr signals, the computation complexity is reduced while maintaining encoding performance.
Solution Approach 2:
The patent changes the computational approach from processing full color difference signals to computing statistical parameters (mean and variance). This parameter transformation simplifies the computational burden while preserving the essential color information needed for accurate encoding.
3Loss of information
If color difference signals are transmitted without processing, then transmission fidelity is maintained, but bit usage increases
Solution Approach 1:
The patent transforms the transmission data from full color difference signals to statistical parameters (mean Co and variance Cg). This parameter change reduces bit usage significantly while maintaining transmission fidelity, as the statistical parameters capture the essential color information needed for accurate reconstruction.
Solution Approach 2:
The patent extracts and transmits only the essential statistical characteristics of color difference signals rather than the complete signal data. This extraction approach reduces bit usage while preserving the fidelity needed for accurate color reconstruction at the decoder.
Data Source
AI summary
An image encoding method includes a color difference signal Cr corresponding to a first pixel and a color difference signal Cb corresponding to a second pixel that are determined based on a pixel value of the first pixel, a pixel value of the second pixel, and pixel values of a third pixel and a fourth pixel that are adjacent to the first pixel. Then, an average value Co of the color difference signal Cr corresponding to the first pixel and the color difference signal Cb corresponding to the second pixel, and a difference value Cg between the color difference signal Cr and the color difference signal Cb are determined.