Color Correction System Using CIE L*a*b* Noise Regulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Digital image processing systems face challenges in color correction, as existing methods often amplify noise, particularly in smaller imaging devices like those used in unmanned aerial vehicles (UAVs), leading to inaccurate color representation and increased noise levels.
Innovation Solution
The implementation of a color correction system that calibrates parameters in the CIE L*a*b* color space to improve accuracy while limiting noise amplification, using a genetic process and a two-step optimization method to avoid local optima, and incorporating a noise evaluation process to assess and reduce noise amplification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If color correction is applied to improve color accuracy, then color representation improves, but noise amplification increases
Solution Approach 1:
The patent transforms the color correction problem from the standard RGB color space to the CIE L*a*b* color space, where the L* channel represents luminance and the a* and b* channels represent color opposites. By performing color correction in this transformed space and applying noise regulation constraints during optimization, the system achieves better color accuracy while controlling noise amplification. The key parameter change is the color space transformation and the introduction of noise regulation parameters in the optimization process.
2Measurement precision
If optimization is performed to improve color correction accuracy, then color precision improves, but the system may get trapped in local optima
Solution Approach 1:
The patent employs a two-step optimization process. First, a genetic algorithm is used to perform preliminary optimization and identify the global optimum or near-global optimum solution. Then, in the second step, gradient-based optimization is applied to refine the solution. This preliminary action of using a global optimization method before local optimization ensures that the system does not get trapped in local optima, thereby improving reliability of convergence.
3Object-generated harmful factors
If noise regulation is incorporated to reduce noise amplification, then noise levels decrease, but computational complexity increases
Solution Approach 1:
The patent incorporates noise regulation by transforming the color values into the CIE L*a*b* color space and applying regularization constraints during the optimization process. The noise regulation is achieved by modifying the objective function to include terms that penalize excessive noise amplification, while still allowing color accuracy improvement. This parameter-based approach integrates noise control directly into the optimization framework without requiring separate complex post-processing steps.
Data Source
AI summary
A computer-implemented method for color correction includes determining a peak signal-to-noise ratio (PSNR) for a noise evaluation image, determining a corrected PSNR for a corrected noise evaluation image, determining a downsampled PSNR for a downsampled noise evaluation image obtained by downsampling the noise evaluation image, determining a downsampled corrected PSNR for a downsampled corrected noise evaluation image obtained by downsampling the corrected noise evaluation image, and determining a noise amplification metric based on the PSNR, the corrected PSNR, the downsampled PSNR, and the downsampled corrected PSNR.