Adaptive Pixel Weighting for Ghost-Free Image De-noising
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image de-noising in the time domain often results in ghosting artifacts or unnatural images due to significant scene changes, necessitating adaptive adjustments to avoid these issues.
Innovation Solution
An image de-noising method that categorizes pixels into low-frequency and high-frequency components, adjusting weights based on scene changes to calculate weighted sums for each, thereby generating de-noised pixels that minimize noise while maintaining image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If time domain image de-noising is performed using simple weighted average, then processing speed is improved and implementation is simplified, but ghosting artifacts and unnatural images occur due to inability to adapt to scene changes
Solution Approach 1:
The patent applies dynamics by making the weighted average adaptive rather than fixed. The weights are dynamically adjusted based on scene change detection and lighting conditions, allowing the de-noising process to adapt to varying conditions while maintaining processing efficiency in the time domain
Solution Approach 2:
The patent changes the parameter of weight values based on scene change detection. When scene changes are detected, the weight allocation is modified to reduce ghosting artifacts. This parameter adjustment allows the system to maintain high processing speed while improving image quality reliability
2Reliability
If adaptive weight adjustment is implemented to handle scene changes, then image quality is improved, but device complexity increases
Solution Approach 1:
The patent segments the image processing into distinct modules: scene change detection module, weight adjustment module, and de-noising module. This segmentation allows each component to be optimized independently, managing overall system complexity while improving image quality through coordinated operation
Solution Approach 2:
The patent introduces scene change detection as an intermediary mechanism that bridges the simple weighted average and the need for adaptive processing. This intermediary detects changes and triggers appropriate weight adjustments, adding complexity only where needed to maintain image quality
3Ease of manufacture
If simple weighted average is used without adaptation, then processing is simplified, but noise removal effectiveness decreases under varying lighting and scene conditions
Solution Approach 1:
The patent changes the weight parameters based on lighting conditions and scene changes. Under stable conditions, simple weighted average is used; under varying conditions, adaptive weight adjustment is applied. This parameter change approach maintains implementation simplicity while significantly improving noise removal effectiveness
Data Source
AI summary
An image de-noising method and an apparatus thereof are disclosed, which includes categorizing a pixel in a current frame into a first low-frequency pixel having a first weight and a first high-frequency pixel having a second weight; categorizing a previous pixel corresponding to the position of the pixel in a previous frame into a second low-frequency pixel having a third weight and a second high-frequency pixel having a fourth weight; adjusting the first weight and the third weight and calculating the weighted sum of the first low-frequency pixel and the second low-frequency pixel, to generate low-frequency pixel data; adjusting the second weight and the fourth weight and calculating the weighted sum of the first high-frequency pixel and the second high-frequency pixel, to generate high-frequency pixel data; and calculating the sum of the low-frequency pixel data and the high-frequency pixel data, to output the de-noised pixel.