Image Density Conversion for Latent Fingerprint Enhancement

Solution Overview

Problem

Existing image enhancement methods struggle to effectively enhance latent fingerprint images with uneven density and background noises, often enhancing noise area boundaries and failing to completely remove background noise, which hampers automation and accuracy in fingerprint matching.

Innovation Solution

An image density conversion method that sets overlapping pixel areas to extract maximum and minimum density values, replacing values to establish local maxima and minima, and linearly converting density values to maintain relative ratios, thereby reconstructing the image to enhance target textures while suppressing noise boundaries.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If Adaptive Contrast Stretch or Adaptive Histogram Equalization is applied to enhance the target texture, then the texture enhancement is improved, but the background noise area boundary is enhanced too much or cannot be removed completely

Engineering Contradiction:

Improvetexture enhancement accuracyVSAvoidbackground noise

Core Design Contradiction:

Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The image is divided into multiple pixel areas with different smoothing levels based on local dynamic range analysis. By segmenting the image into regions requiring different enhancement intensities, the method enhances target textures while avoiding excessive enhancement of noise boundaries.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Different smoothing levels are applied to different pixel areas based on their local characteristics. The method calculates local dynamic ranges and assigns appropriate smoothing levels to each area, ensuring that enhancement is adapted to local image properties rather than applying uniform enhancement.

Inventive Principle:

Principle #3Local quality

2Ease of operation

If a certain appropriate reference area is provided around a target pixel for density conversion, then the conversion is simplified, but the background noise area is enhanced too much or cannot be removed completely

Engineering Contradiction:

Improvedensity conversion simplicityVSAvoidbackground noise

Core Design Contradiction:

Ease of operationVSObject-affected harmful factors

Solution Approach 1:

The image is divided into multiple pixel areas with different smoothing levels based on local dynamic range analysis. By segmenting the image into regions requiring different enhancement intensities, the method enhances target textures while avoiding excessive enhancement of noise boundaries.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The smoothing level parameter is dynamically changed based on local dynamic range characteristics. By adjusting the smoothing level according to local image properties, the method achieves adaptive enhancement that preserves target textures while suppressing noise.

Inventive Principle:

Principle #35Parameter changes

3Adaptability or versatility

If area division is performed based on texture analysis to determine smoothing levels, then the enhancement is adapted to local characteristics, but the enhancement result deteriorates if area division is not processed as expected

Engineering Contradiction:

Improvelocal adaptation capabilityVSAvoidenhancement result stability

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The smoothing level parameter is dynamically changed based on local dynamic range characteristics. By adjusting the smoothing level according to local image properties, the method achieves adaptive enhancement that preserves target textures while suppressing noise.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The method performs preliminary analysis of local dynamic ranges before applying enhancement. By pre-calculating the characteristics of each pixel area and determining appropriate smoothing levels in advance, the method ensures reliable and consistent enhancement results.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS7961968B2Image density conversion method, image enhancement processor, and program thereof

Publication Date: 2011.06.14 NEC CORP

AI summary

Areas, allowed to overlap, are set for respective pixels in an image, and a maximum density value and a minimum density value, among inherent density values of the pixels in each area, are extracted and specified as maximum and minimum density values common to the pixels in the area. Among the density values specified from the overlapping areas, the largest minimum density value is set as a local minimum value and the smallest maximum density value is set as a local maximum value to thereby generate a minimum density plane and a maximum density plane including the pixels having the values. The density values of the pixels at the corresponding positions in the minimum density plane and the maximum density plane are corresponded to the common minimum and maximum values set separately, to thereby linearly convert the whole. Thereby, the pixel density values of the original image are linearly converted simultaneously.