CMOS Image Sensor Noise Removal Using Reference Pixels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In CMOS image sensors with a column parallel structure, row-to-row noise caused by power supply to pixels is not effectively removed, leading to fluctuations and shading in pixel data due to the inclusion of this noise in the output signal.
Innovation Solution
The integration of dummy or reference pixels in row lines to sense common mode noise, combined with a differential readout block operation to remove both row-to-row noise and comparator-induced noise, utilizing a pixel signal processing apparatus that demultiplexes and calculates differences between reference values and pixel signals to produce noise-free output.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a column parallel structure is used with differential mode comparators, then comparator-induced noise is effectively removed, but row-to-row noise from pixel power supply is not removed, causing row-to-row fluctuation and shading
Solution Approach 1:
The pixel array is segmented into multiple types of pixels: active pixels for image capture and reference pixels (including first reference pixels and second reference pixels) for noise sensing. This segmentation allows the system to separately measure and compensate for different noise types. The readout block is also segmented to include multiple comparators that can process different pixel types independently.
Solution Approach 2:
Reference pixels serve as intermediary elements that sense the row-to-row noise caused by pixel power supply. These reference pixels are positioned at specific locations (first and second reference pixels) and provide reference signals that are used to compensate for the noise in the active pixel signals through the pixel signal processing apparatus.
2Measurement precision
If reference pixels are added to sense common mode noise, then row-to-row noise is removed, but device complexity increases
Solution Approach 1:
Reference pixels are strategically positioned at specific locations within the pixel array (first reference pixels and second reference pixels at different positions) to sense different aspects of common mode noise. This localized placement allows efficient noise sensing without requiring reference pixels throughout the entire array, thereby limiting the increase in device complexity.
Solution Approach 2:
The reference pixels serve multiple functions: they sense row-to-row noise, provide reference signals for compensation, and enable the system to handle different types of noise (comparator-induced and pixel-power-supply-induced). This multi-functionality reduces the need for separate dedicated circuits for each noise type.
3Measurement precision
If pixel signal processing apparatus performs linear operation and difference calculation, then both types of noise are removed, but processing time increases
Solution Approach 1:
The reference pixels sense and store reference signals in advance before the main image processing occurs. The pixel signal processing apparatus then uses these pre-sensed reference signals to compensate for noise in the active pixel signals, performing the difference calculation and linear operation efficiently without requiring time-consuming real-time measurements.
Data Source
AI summary
A CMOS image sensor may include an active pixel sensor suitable for generating a pixel signal corresponding to incident light, a reference pixel array suitable for shielding incident light and generate a reference value, a readout circuit suitable for comparing the pixel signal and the reference value with a ramp signal and for removing a first noise of the pixel signal in a common mode, and a pixel signal processing circuit suitable for performing a linear operation on an output signal from the readout circuit and the reference value, and for removing a second noise of the pixel signal in the common mode.