Fingerprint Core Extraction via Skeleton Loop Score and Density Inversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current fingerprint matching systems face challenges in accurately extracting fingerprint cores from low-quality latent fingerprints and high-quality inked fingerprints, particularly due to noise and complex shapes, which affects matching accuracy and increases computational load.
Innovation Solution
A fingerprint core extraction device and method that generates a skeleton image from ridgeline shapes, performs endpoint processing, and calculates loop scores to determine the fingerprint core, using density inversion to enhance core detection accuracy and stability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional core extraction methods using ridgeline direction are used, then stable core extraction is achieved, but detailed position detection cannot be expected
Solution Approach 1:
The patent segments the skeleton image processing into distinct stages: extracting bifurcations, performing endpoint processing, generating density inverted images, and detecting loop-like vertices. This segmentation allows each stage to focus on specific aspects of core detection, improving overall precision without overwhelming complexity in any single step.
Solution Approach 2:
The patent applies density inversion to the skeleton image, transforming the image so that dark regions become light and vice versa. This inversion enhances the visibility of loop-like structures and core regions, enabling more precise detection of fingerprint cores while maintaining manageable processing complexity through the systematic approach.
2Reliability
If skeleton images with noise or complex shapes are used, then core detection becomes unreliable, but the patent achieves stable detection through endpoint processing
Solution Approach 1:
The patent performs preliminary actions by extracting bifurcations and executing endpoint processing on the skeleton image before detecting loop-like vertices. This preliminary processing removes noise and simplifies complex shapes, creating a cleaned-up skeleton structure that enables reliable core detection even from noisy or complex original images.
Solution Approach 2:
The patent converts the harmful effects of noise and complex shapes into beneficial outcomes by using density inversion. The inversion process transforms noisy regions into distinguishable patterns, and complex shapes are converted into identifiable loop-like structures, thereby improving detection reliability rather than being hindered by the original image qualities.
3Productivity
If minutia extraction is used for fingerprint matching, then matching can be performed, but sufficient minutiae cannot be extracted from latent fingerprints of small region
Solution Approach 1:
The patent extracts the essential core information from the fingerprint skeleton image, focusing on loop-like vertices and their geometric properties. By taking out only the critical core features rather than attempting to extract all minutiae, the system achieves effective fingerprint matching even from latent fingerprints with limited regions, overcoming the shortage of extractable minutiae.
Solution Approach 2:
The patent changes the parameter of focus from traditional minutiae (endpoints and bifurcations) to core position and loop-like vertex geometry. This parameter change enables the system to extract sufficient matching information from latent fingerprints with small regions, as the core and its geometric properties provide adequate characteristics for matching without requiring numerous minutiae.
Data Source
AI summary
To extract a fingerprint core for alignment of a fingerprint image for fingerprint matching. The present invention comprises: a core extraction unit which has an endpoint processing function of generating an endpoint image by extracting bifurcations in skeleton image and subjecting the respective bifurcations to endpoint processing; and a skeleton image generation/extraction unit which extracts skeletons of a density-inverted image formed by inverting the density colors of the generated end point image from white to black and vice versa, and performs processing for generating the skeleton image, the core extraction unit being provided with a core determination function of specifying a skeleton pixel having the highest value of a loop score indicating the degree of a bend in a skeleton shape in the density-inverted image, and defining this skeleton pixel as the finger print core.