Video Denoising via Block Variance Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video denoising algorithms face challenges in accurately estimating noise levels, leading to flickering or blurring issues and inefficient denoising performance, especially in videos captured by smartphones with limited aperture and sensor size, due to high time complexity and uniform denoising strength across all pixel points without considering brightness variations.
Innovation Solution
A video denoising method that divides each frame into sub-image blocks, calculates block variance, determines an average variance to estimate the noise level, and selects denoise strength and noise characteristic curves matching the noise level, incorporating spatial and temporal variances and smoothing noise levels over multiple frames to adaptively denoise frames using algorithms like BM3D with brightness-aware scaling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If adaptive denoising algorithm dynamically adjusts denoising parameters by estimating noise level, then image quality is improved, but noise level estimation accuracy deteriorates
Solution Approach 1:
The image is divided into multiple sub-image blocks, and each block's variance is calculated independently. This segmentation allows the algorithm to estimate noise levels locally rather than globally, improving the accuracy of noise level estimation while maintaining image quality through adaptive denoising parameters.
Solution Approach 2:
The patent applies different denoising parameters to different regions of the image based on local noise level estimates. By calculating variance for each sub-image block separately, the algorithm adapts to local variations in noise characteristics, thereby improving both noise level estimation accuracy and overall image quality.
2Adaptability or versatility
If variance calculation is performed on sub-image blocks to estimate noise level, then denoising adaptability is improved, but details are misidentified as noise
Solution Approach 1:
The patent dynamically adjusts denoising parameters based on the calculated variance of each sub-image block. By making the denoising process adaptive rather than static, the algorithm can distinguish between actual noise and important image details, reducing detail loss while maintaining denoising adaptability.
Solution Approach 2:
The algorithm changes denoising parameters based on the variance values calculated from sub-image blocks. By adjusting parameters dynamically according to local variance characteristics, the system achieves better adaptability without misidentifying important details as noise.
3Manufacturing precision
If video block-matching and 3D filtering (VBM3D) or 4D filtering (VBM4D) algorithms are used, then video denoising performance is improved, but time complexity and hardware resource cost increase
Solution Approach 1:
The patent extracts only the essential variance calculation from complex video block-matching algorithms. By focusing on calculating the variance of sub-image blocks rather than performing full block-matching and 3D/4D filtering, the system achieves acceptable denoising performance with significantly reduced time complexity and hardware resource requirements.
Solution Approach 2:
The patent uses a simpler, computationally cheaper variance-based noise estimation method instead of expensive VBM3D or VBM4D algorithms. This disposable approach calculates only what is necessary (variance of sub-blocks) to achieve denoising, avoiding the heavy computational burden of more complex algorithms while maintaining practical performance.
4Ease of operation
If uniform denoising strength is applied to all pixel points in a frame, then processing simplicity is improved, but denoising effectiveness deteriorates due to ignoring brightness variations
Solution Approach 1:
The patent applies different denoising strengths to different regions of the image based on local variance calculations. Instead of using a uniform denoising strength, the system adapts the denoising parameters to local characteristics, thereby improving denoising effectiveness while maintaining reasonable processing simplicity through automated local adaptation.
Data Source
AI summary
The present disclosure discloses a video denoising method and device, and a computer-readable storage medium. The video denoising method includes: dividing each video frame in an input video frame sequence into sub-image blocks, and calculating a block variance of each of the sub-image blocks; calculating an average variance of the sub-image blocks in a current video frame according to the calculated block variances, determining a noise level of the current video frame according to the average variance, selecting a denoise strength and a noise characteristic curve that match the noise level; and denoising the current video frame according to the denoise strength and the noise characteristic curve.