Image Noise Reduction via Spectral Transforms on Row Column Means
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image noise reduction techniques are inefficient in detecting and removing spatially correlated temporal row and column noise, particularly in embedded systems with limited computational power, as they require processing of individual pixel values and struggle to distinguish between scene-based information and noise.
Innovation Solution
Applying spectral transforms to row and column values of captured image frames to selectively reduce noise, using first and second spectral transforms to differentiate noise from scene information and adjust coefficients to filter out noise, thereby reducing spatially correlated temporal noise in a computationally efficient manner.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional pixel-based noise reduction techniques are used, then noise detection capability is improved, but computational complexity increases significantly
Solution Approach 1:
The patent segments the image data by extracting only row and column mean values from each frame, rather than processing all pixel values. This segmentation reduces the data volume from millions of pixels to just a few thousand row and column means, enabling efficient noise detection while maintaining the ability to identify spatially correlated noise patterns.
Solution Approach 2:
The patent extracts only the essential features (row and column mean values) needed for noise detection, discarding redundant pixel-level information. By taking out only the necessary statistical representations of each row and column, the system achieves effective noise detection with minimal computational resources.
2Measurement precision
If individual pixel processing is performed, then noise discrimination accuracy is improved, but processing speed decreases
Solution Approach 1:
The patent segments the processing task by operating on aggregated row and column mean values rather than individual pixels. This segmentation maintains noise discrimination accuracy because spatially correlated noise manifests clearly in these aggregated statistics, while simultaneously improving processing speed by reducing the data volume by several orders of magnitude.
3Reliability
If comprehensive pixel analysis is performed, then noise identification reliability is improved, but memory requirements increase
Solution Approach 1:
The patent extracts only the essential statistical features (row and column means) needed for reliable noise identification, storing and processing just these aggregated values rather than maintaining copies of entire image frames or individual pixel data. This extraction approach ensures reliable noise detection while minimizing memory consumption to levels suitable for embedded systems.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Various techniques are provided for reducing noise in captured image frames. In one example, a method includes determining row values for image frames comprising scene information and noise information. The method also includes performing first spectral transforms in a first domain on corresponding subsets of the row values to determine first spectral coefficients. The method also includes performing second spectral transforms in a second domain on corresponding subsets of the first spectral coefficients to determine second spectral coefficients. The method also includes selectively adjusting the second spectral coefficients. The method also includes determining row correction terms based on the adjusted second spectral coefficients to reduce the noise information of the image frames. Additional methods and systems are also provided.