Image Sensor Fixed Pattern Noise Removal for Saturation Pixels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image sensing devices suffer from noise issues due to varying pixel sensitivity and fixed pattern noise, particularly in saturation areas, which affect image resolution.
Innovation Solution
An image processing method and device that classifies pixel values into saturation and non-saturation images, calculates separate fixed pattern noises for each, and combines these to generate a final noise to be subtracted from the raw image, thereby improving resolution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If fixed pattern noise is removed using a single uniform method for all pixels, then the processing is simple, but the noise removal accuracy deteriorates due to varying pixel sensitivity
Solution Approach 1:
The patent segments the pixel array into multiple sensitivity groups based on pixel sensitivity characteristics. Each group is processed with its own fixed pattern noise map, allowing accurate noise removal tailored to each group's sensitivity profile while maintaining manageable processing complexity through systematic classification.
Solution Approach 2:
The patent applies local quality by using different fixed pattern noise maps for different pixel sensitivity groups. Each region (sensitivity group) receives customized noise removal parameters matching its specific characteristics, improving overall noise removal accuracy without requiring completely separate processing for every individual pixel.
2Measurement precision
If pixel sensitivity variation is compensated by individual calibration, then noise removal accuracy is improved, but device complexity increases
Solution Approach 1:
The patent reduces device complexity by segmenting pixels into a limited number of sensitivity groups rather than calibrating each pixel individually. This grouping approach maintains high noise removal accuracy while significantly reducing the computational burden and processing complexity compared to full individual calibration.
Solution Approach 2:
The patent creates universal fixed pattern noise maps for each sensitivity group that can be applied to multiple pixels sharing similar sensitivity characteristics. This multi-functional approach allows a single noise map to serve multiple pixels, reducing overall processing complexity while maintaining accuracy for each pixel type.
3Measurement precision
If saturation pixels are processed differently from non-saturation pixels, then noise removal accuracy in saturation areas is improved, but processing complexity increases
Solution Approach 1:
The patent applies local quality by creating specific fixed pattern noise maps for saturation pixels within each sensitivity group. This targeted approach improves noise removal accuracy in saturation areas by accounting for their unique characteristics, while the overall systematic framework keeps processing complexity manageable.
Solution Approach 2:
The patent segments saturation pixels as a distinct category within the sensitivity group framework. This segmentation allows specialized noise removal processing for saturation pixels without requiring completely separate handling, balancing improved accuracy with controlled processing complexity through hierarchical classification.
Data Source
AI summary
An image processing device and an image processing method are provided. The image processing device includes an image sensor configured to sense and output a raw image and an image signal processor configured to compare pixel values of the raw image with a reference level to classify a saturation color image and a non-saturation color image, calculate a first fixed pattern noise from the saturation color image, calculate a second fixed pattern noise from the non-saturation color image, generate a final fixed pattern noise by combining the first fixed pattern noise with the second fixed pattern noise, and output a revised image by subtracting the final fixed pattern noise from the raw image.