CMOS Image Sensor Noise Reduction via Line-Optical Black Pixels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
CMOS image sensors suffer from noise generation that degrades image quality, leading to lower resolution compared to CCD image sensors, and there is a need for an image sensor that can produce higher quality images while reducing noise.
Innovation Solution
An image sensor with a pixel array that includes active pixels and line-optical black (L-OB) pixels arranged in a matrix, where dark-level offset signals are simultaneously output from L-OB pixels in different rows and columns to reduce noise, and an analog-to-digital converter (ADC) block digitizes these signals for noise reduction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If CMOS image sensor is used, then manufacturing cost is reduced and device size is decreased, but image quality is degraded due to noise generation
Solution Approach 1:
The pixel array is segmented into active pixels and L-OB pixels, where L-OB pixels are specifically dedicated to generating dark-level offset signals for noise reduction. This segmentation allows the system to maintain the cost-effective CMOS structure while introducing specialized pixels that improve image quality by providing reference signals for subtracting row noise and pixel intrinsic noise.
Solution Approach 2:
Dark-level offset signals generated by L-OB pixels serve as an intermediary mechanism to eliminate noise. These offset signals are subtracted from the pixel signals to remove row noise and pixel intrinsic noise, thereby improving image quality without changing the fundamental CMOS sensor structure or manufacturing process.
2Reliability
If multiple L-OB pixels simultaneously output dark-level offset signals, then noise reduction is improved, but device complexity increases
Solution Approach 1:
Multiple dark-level offset signals from different L-OB pixels are merged and averaged to produce a single representative offset signal. This averaging process improves noise reduction effectiveness by combining multiple measurements, while the systematic approach to merging keeps the processing complexity manageable through regular algorithms.
Data Source
AI summary
An image sensor includes a pixel array including at least one active pixel and at least one line-optical black (L-OB) pixel arranged in a matrix including first to nth rows and first to mth columns, the pixel array configured to output a pixel signal and a dark-level offset signal in units of columns during a read-out operation in one of the first to nth rows; a row driver configured to output a selection control signal to the first to nth rows; and an analog-to-digital converter (ADC) block configured to digitize the pixel signal and the dark-level offset signal. In the pixel array, a dark-level offset signal is simultaneously output from an L-OB pixel in another row during the read-out operation in one of the first to nth rows. Here, ‘n’ and ‘m’ each denote an integer that is equal to or greater than ‘2’.