Image Sensor Column Offset Correction Scaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
CMOS image sensors suffer from column fixed pattern noise due to mismatches in column output circuitry offsets and gains, which become more noticeable in low light conditions and are amplified by image gain, degrading image quality.
Innovation Solution
A method to determine and scale column offset corrections using dark reference pixels, where initial corrections are calculated at a base gain level and updated based on changes in gain levels, reducing the need for repeated dark signal readings and compensating for outlier signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If gain is increased to make images brighter in low light conditions, then image brightness is improved, but column fixed pattern noise is amplified and becomes more noticeable
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing offset correction values at multiple gain levels before image capture. When a gain change occurs, the system retrieves and applies the pre-computed correction values corresponding to the new gain level, avoiding the need to recompute corrections after gain changes. This proactive preparation eliminates column fixed pattern noise at the new gain level without requiring additional dark signal readings.
Solution Approach 2:
The patent implements parameter changes by computing offset corrections at multiple discrete gain levels (e.g., 1x, 2x, 4x, 8x) and selecting the appropriate correction set based on the current gain level. The system detects gain level changes and switches between pre-computed correction sets corresponding to different gain parameters, ensuring noise compensation is optimized for the actual gain being used without requiring continuous recalibration.
2Measurement precision
If dark signal readings are repeatedly taken to update offset corrections for each gain level change, then noise compensation accuracy is improved, but frame overhead and processing time increase
Solution Approach 1:
The system performs preliminary dark signal readings and offset correction computations at multiple gain levels during initialization or calibration phases, storing these correction values for later use. When operation begins, the pre-computed corrections are already available, eliminating the need to perform time-consuming dark signal readings and computations during actual image capture sequences.
Solution Approach 2:
Instead of repeatedly measuring dark signals for each gain change, the system changes the parameter of which correction set is applied based on the current gain level. The system detects the active gain level and selects the corresponding pre-computed correction set from memory, achieving accurate noise compensation without repeated measurements.
3Manufacturing precision
If offset corrections are updated for every gain level change, then image quality across varying gain levels is improved, but device complexity and processing requirements increase
Solution Approach 1:
The patent manages complexity by organizing corrections into discrete parameter sets corresponding to specific gain levels. The system simply detects which gain level is active and applies the matching correction set, avoiding complex real-time adaptation algorithms. This parameter-based approach maintains high image quality across gain levels while keeping the correction mechanism relatively simple and efficient.
Data Source
AI summary
An image sensor includes multiple photoactive pixels and multiple dark reference pixels typically arranged in rows and columns to form a pixel array. A dark signal is read out from a given number of dark reference pixels in each column at a first gain level. An initial column offset correction is determined for one or more columns in the pixel array using respective dark signals read out at the first gain level. The initial column offset corrections are repeatedly scaled in response to each detected change to a different gain level. The column offset corrections can be scaled based on an amount of change between each respective different gain level and the first gain level.