Adaptive Wavelet Denoising for Image Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image and video compression techniques face challenges in efficiently denoising images before encoding, especially when noise characteristics are unknown or time-varying, leading to complex denoising processes with high computational load and potential for under- or over-denoising.
Innovation Solution
The use of adaptive wavelet denoising techniques, specifically wavelet shrinkage-based methods, which generate 2-D wavelet coefficients at multiple levels, identify flat regions, determine threshold variances, and apply thresholding to remove noise, resulting in a denoised image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If traditional denoising techniques are used, then noise removal is achieved, but computational load increases and processing complexity worsens
Solution Approach 1:
The image is divided into multiple blocks, and each block is processed independently to identify flat regions and determine threshold variances. This segmentation allows the denoising process to focus computational resources only where needed (in flat regions) rather than processing the entire image uniformly, thereby reducing overall computational load and processing complexity while maintaining effective noise removal.
Solution Approach 2:
The patent applies different denoising operations to different regions of the image based on local characteristics. Flat regions identified through variance analysis receive thresholding operations, while non-flat regions are preserved with minimal processing. This local quality approach ensures that noise removal is applied selectively where it is most beneficial, reducing unnecessary computational complexity in regions where denoising would not improve quality.
2Object-affected harmful factors
If aggressive denoising is applied, then noise reduction improves, but image quality deteriorates due to under- or over-denoising
Solution Approach 1:
The patent employs adaptive threshold variance determination that dynamically adjusts denoising parameters based on local image characteristics. By calculating variance within flat regions and using this to determine appropriate thresholds, the system adapts to different noise levels and image regions, preventing both under-denoising (which would leave noise artifacts) and over-denoising (which would blur important details). This dynamic adaptation maintains image quality while achieving effective noise reduction.
Solution Approach 2:
The variance calculation of flat regions provides feedback about the actual noise characteristics in different parts of the image. This feedback is used to adjust the threshold variance and subsequent thresholding operations, creating a closed-loop system that automatically adapts to the actual noise conditions rather than applying fixed, aggressive parameters. This feedback mechanism ensures optimal noise reduction while preserving image quality.
3Object-affected harmful factors
If complex denoising processes are used, then noise removal effectiveness improves, but processing time increases
Solution Approach 1:
The patent performs preliminary identification of flat regions through variance analysis before applying the computationally intensive thresholding operations. By pre-identifying which regions are suitable for denoising (flat regions with low variance) and which should be preserved (non-flat regions with higher variance), the system avoids unnecessary processing in regions where denoising would not be beneficial. This preliminary action significantly reduces processing time while maintaining noise removal effectiveness in the regions that require it.
Data Source
AI summary
Image data is processed for noise reduction before encoding and subsequent decoding. For an input image in a spatial domain, two-dimensional (2-D) wavelet coefficients at multiple levels are generated. Each level includes multiple subbands, each associated with a respective subband type in a wavelet domain. For respective levels, a flat region of a subband is identified, which flat region includes blocks of the subband having a variance no higher than a first threshold variance. A flat block set for the subband type associated with the subband is identified, which includes blocks common to respective flat regions of the subband. A second threshold variance is determined using variances of the flat block set, and is then used for thresholding at least some of the 2-D wavelet coefficients to remove noise. After thresholding, a denoised image is generated in the spatial domain using the levels.