Digital Image Enhancement via Frequency Band Decomposition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Digital images, particularly medical images, face challenges with excessive noise and poor contrast, leading to reduced visibility of objects of interest, as existing image processing techniques have limited success in enhancing images without adversely affecting patient health or introducing unwanted noise enhancement.
Innovation Solution
A method and system that decompose digital images into frequency bands, suppress noise, and enhance contrast by computing representative signal values from neighborhood pixel contributions, applying lookup tables to selectively enhance signals and suppress noise, and recombining the images to produce an enhanced image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If patient exposure or dose is increased to improve signal-to-noise ratio, then image visibility is improved, but patient health is adversely affected
Solution Approach 1:
The patent extracts and removes noise components from the image signal through digital processing techniques, separating the useful signal from harmful noise without requiring increased patient exposure. This allows improvement of signal-to-noise ratio while maintaining original dose levels, thus resolving the contradiction between measurement precision and patient health protection.
2Measurement precision
If simple interactive adjustments of contrast and brightness are applied, then contrast problem is addressed, but acceptable contrast cannot be produced throughout entire image simultaneously and time is consumed
Solution Approach 1:
The patent segments the image into multiple frequency bands or resolution grids, allowing independent processing of different spatial frequencies. This enables automatic optimization of contrast at multiple scales simultaneously without requiring manual interactive adjustments, thus achieving acceptable contrast throughout the entire image while reducing time consumption.
Solution Approach 2:
The patent applies non-linear lookup tables and transforms to modify pixel values based on their original values and neighborhood characteristics. This automatic parameter transformation optimizes contrast distribution across the entire image dynamically, eliminating the need for manual interactive adjustments and achieving both good contrast and time efficiency.
3Measurement precision
If non-linear lookup tables are applied to each sub-band image to enhance contrast, then contrast enhancement is achieved, but noise is enhanced as well as signal
Solution Approach 1:
The patent applies different processing characteristics to different regions and frequency bands of the image. By analyzing local neighborhood values and applying adaptive lookup tables that depend on representative values of neighborhoods, the system enhances contrast in regions where it is needed while preserving noise characteristics in other regions, thus achieving selective contrast enhancement without uniform noise amplification.
Data Source
AI summary
A system and method for enhancing digital images is described. A digital image is transformed into a series of decomposed images in frequency bands, or different resolution grids. A decomposed image is noise suppressed and contrast enhanced. Representative value of signal at each pixel is computed based on contributions to signals from pixels in a neighborhood of the pixel. Lookup tables are applied to pixel values to selectively enhance signal in a predetermined range of signal strength. Another set of lookup tables are applied to pixel values to suppress noise components contained therein. Optionally, operations are applied to decomposed images to suppress quantum noise, enhance object edges or enhance global contrast, among others. These decomposed images, after signal enhancement and noise suppression, are then recombined to result in an enhanced image.