Image Processing Device Offset Correction for CMOS Noise
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In imaging elements, such as CMOS, the decrease in pixel and read circuit size leads to reduced sensitivity and increased noise, including dark current shot noises and random noises, making it difficult to accurately detect and correct defective pixels, especially when adjacent defective blocks occur.
Innovation Solution
An image processing device that includes an acquisition unit for acquiring image data from sharing blocks with multiple adjacent pixels sharing a common read circuit, an offset calculator to detect offset components, and a corrector to calculate and apply correction amounts to pixel values within the sharing block, thereby improving sensitivity and noise reduction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of moving object
If pixel size and read circuit size are diminished to reduce area and increase aperture ratio, then sensitivity is improved, but noise increases and sensitivity actually deteriorates
Solution Approach 1:
The imaging element is divided into multiple sharing blocks, where each block contains multiple pixels that share a common read circuit. This segmentation allows for independent noise correction processing at the block level, enabling effective suppression of read circuit noise while maintaining the area benefits of sharing structures.
Solution Approach 2:
An offset calculator and corrector are introduced as intermediary components that process pixel values within each sharing block. These intermediaries detect offset components caused by read circuit noise and apply corrections, thereby mediating between the area-efficient sharing structure and the quality requirements of the output signal.
2Measurement precision
If sharing block size is increased to improve noise correction accuracy, then measurement precision improves, but device complexity increases
Solution Approach 1:
The image data is segmented into multiple sharing blocks that are processed independently. This segmentation enables parallel processing of noise correction for each block, reducing the overall computational complexity compared to processing the entire image as a single unit while still achieving sufficient correction accuracy within each block.
Solution Approach 2:
The noise correction is applied partially by processing only the offset component within each sharing block rather than attempting to correct all types of noise in the entire image. This partial action approach achieves practical correction accuracy while maintaining manageable processing complexity.
3Productivity
If multiple sharing blocks are processed independently to reduce processing complexity, then productivity improves, but measurement precision deteriorates due to boundary effects
Solution Approach 1:
Adjacent sharing blocks are processed in combination, where the offset calculation for one block utilizes pixel value information from neighboring blocks. This merging approach ensures continuity and consistency at block boundaries, eliminating precision deterioration while maintaining the productivity benefits of block-based parallel processing.
Data Source
AI summary
An image processing device includes: an acquisition unit configured to acquire image data generated by an imaging element including sharing blocks, each including multiple adjacent pixels configured to generate signals according to amounts of light received externally and one read circuit; an offset calculator configured to, based on the acquired image data and using pixel values of pixels in a sharing block of interest that is one of the multiple sharing blocks and pixel values of pixels outside the sharing block of interest, detect an offset component of a pixel value occurring in the sharing block of interest; and a corrector configured to calculate a correction amount to correct a pixel value of at least one pixel in the sharing block of interest based on the offset component and use the correction amount to correct a pixel value of at least one pixel in the sharing block of interest.