Image Sensor Color Filter Array Edge Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image sensors with Bayer patterns face challenges in reducing noise at the edges of captured images due to non-adjacent color filter pixels, leading to noise amplification during lens shading correction, especially in low illuminance environments, which deteriorates image quality.
Innovation Solution
The use of large patterns such as tetra, nona, or hexa-deca patterns at the edges of the color filter array allows adjacent color filter pixels, enabling more efficient noise reduction and improving image quality by employing a binning unit to selectively perform binning based on light amounts in the image frame.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a Bayer pattern is used in the color filter array, then the image sensor can capture color information, but noise is amplified at the edges of captured images due to non-adjacent color filter pixels
Solution Approach 1:
The patent applies different color filter patterns to different regions of the pixel array. Specifically, a first color filter pattern (Bayer pattern) is applied to a first region, while a second color filter pattern (tetra pattern with adjacent same-color pixels) is applied to a second region, particularly at the edges. This local differentiation allows the edge regions to benefit from reduced noise amplification while the center region maintains standard color filtering.
Solution Approach 2:
The pixel array is divided into multiple regions with different color filter patterns. The patent segments the array such that central pixels use one pattern while edge pixels use another pattern, allowing optimized noise reduction at edges without compromising overall color imaging capability.
2Reliability
If lens shading correction is performed to correct illumination non-uniformity, then image quality improves, but noise is amplified more significantly in low illuminance environments
Solution Approach 1:
By applying the tetra pattern specifically to edge regions where lens shading effects are most pronounced, the patent provides localized noise reduction exactly where lens shading correction causes the most noise amplification, thereby mitigating the harmful effect during correction processing.
3Measurement precision
If the number of pixels is increased to improve resolution, then image detail is enhanced, but noise reduction becomes more difficult at the edges
Solution Approach 1:
The patent maintains high pixel density across the entire array for maximum resolution, but applies a specialized tetra pattern specifically to edge pixels. This allows the bulk of the array to maintain high resolution with standard Bayer pattern, while edge pixels are reconfigured to provide noise reduction, thus preserving both resolution and reducing edge noise.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This configuration reduces noise effectively and enhances image quality by maintaining signal integrity and reducing noise amplification during lens shading correction, particularly in low-light conditions.
Implementation Method 1
Each of the pixels includes a photodiode. The photodiode converts incident light into an electrical signal.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An image device is provided. The image device includes a pixel array comprising a first pixel array portion of a part of the pixel array and a second pixel array portion of another part of the pixel array that outputs an image; and a color filter array disposed over the pixel array, the color filter array comprising a first color filter pattern on the first pixel array portion and a second color filter pattern on the second pixel array portion, which are comprising a plurality of different color filters in each of the first pixel array portion and the second pixel array portion,, wherein the plurality of the color filters in the first color filter pattern is arranged in an array of N ∗ N and the plurality of the color filters in the second color filter pattern is arranged in an array of M ∗ M, wherein N and M are a natural number of 2 or more, and the N and M is different each other, and wherein a number of red (R) color filters included in the first color filter pattern is equal to a number of R color filters included in the second color filter pattern.