On-Chip FPN Correction Using Dual Column Sample-and-Hold Subtraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for correcting fixed pattern noise in imagers require large memory to store correction coefficients, which is inefficient and resource-intensive.
Innovation Solution
A method involving dual column readout using sample and hold circuits to subtract dark signals from light-related signals, eliminating the need for large memory storage by performing the subtraction in real-time using a single subtraction unit.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If dark signal levels for each pixel are stored in memory for FPN correction, then fixed pattern noise is corrected, but large memory capacity is required
Solution Approach 1:
The pixel array is divided into multiple columns, and dark signal correction is performed column-by-column using separate sample-and-hold circuits for each column. This segmentation allows the system to process and correct FPN in smaller units sequentially, eliminating the need to store all dark signal levels simultaneously in a large memory.
Solution Approach 2:
Dark signal levels are measured and stored in sample-and-hold circuits before actual image acquisition. By performing this preliminary measurement and storage in distributed column-specific circuits rather than centralized memory, the system prepares correction data in advance without requiring large memory capacity during operation.
2Reliability
If dark signal levels are stored and subtracted from light signals, then fixed pattern noise is reduced, but additional signal processing complexity increases
Solution Approach 1:
The dark signal subtraction is merged with the regular signal readout process. The sample-and-hold circuits for dark signals are integrated with the column readout circuitry, and the subtraction operation is combined with the existing signal processing pipeline, reducing overall system complexity compared to separate correction systems.
Solution Approach 2:
Each column's sample-and-hold circuit automatically holds its own dark signal level and performs self-subtraction from the corresponding light signal during readout. This self-service approach eliminates the need for complex centralized control and processing logic, simplifying the overall signal processing architecture.
Data Source
AI summary
A circuit and method for correcting pixel output signals for fixed pattern noise. Pixels in a selected row of pixels are read after an integration period and the resulting signals are stored in a first sample and hold circuit for each column. The pixels in the selected row are then reset and immediately read again and the resulting signals are stored in a second sample and hold circuit for each column. The signals in the second sample and hold circuits are subtracted from the signals in the first sample and hold circuits to produce signals related to the light seen by the pixels in the selected row corrected for fixed pattern noise. The output of the first sample and hold circuits and second sample and hold circuits can be connected to a subtraction unit and sequentially activated so that a single subtraction unit is required for the entire imager. The output of the subtraction unit can connected to a buffer thereby storing signals corrected for fixed pattern noise in the buffer using only a single subtraction unit and avoiding the need for a large memory to store dark pixel signals.