Fingerprint Recognition via Block Partitioning and Frequency Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing fingerprint recognition technologies face challenges in accurately authenticating users due to deformations in fingerprint images caused by pressure variations during sensing, and the need for efficient processing of partial fingerprint images.
Innovation Solution
The method involves partitioning an input partial fingerprint image into blocks and comparing these blocks to enrolled partial images using frequency-based matching, while accounting for rotation, translation, and scale differences between the blocks and enrolled images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If fingerprint images are captured under pressure variations, then sensing capability is improved, but image deformation occurs reducing recognition accuracy
Solution Approach 1:
The fingerprint image is divided into multiple blocks, and each block is independently processed to calculate rotation angles and translations. This segmentation allows the system to handle deformations locally in each block rather than requiring perfect global alignment, thereby maintaining recognition accuracy despite pressure-induced deformations.
Solution Approach 2:
The patent transforms the fingerprint image from spatial domain to frequency domain using Fourier Transform, then applies log-polar transformation to convert rotational and scaling variations into translational shifts in the frequency domain. This parameter transformation makes the matching process invariant to rotation and scale changes caused by pressure variations.
2Loss of time
If partial fingerprint images are used for recognition, then processing speed is improved, but matching accuracy deteriorates due to rotation, translation, and scale differences
Solution Approach 1:
By applying Fourier Transform followed by log-polar transformation, the patent converts rotation, translation, and scale parameters into simpler translational parameters in the frequency domain. This allows efficient comparison of partial fingerprint blocks while maintaining accuracy despite geometric transformations.
Solution Approach 2:
The patent moves the fingerprint matching problem from the spatial domain to the frequency domain, where geometric transformations (rotation, scaling, translation) manifest as simpler parameter changes. This dimensional transformation enables faster and more accurate matching of partial fingerprint images.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A fingerprint recognition method includes receiving an input partial image corresponding to a partial image of a fingerprint of a first user; partitioning the input partial image into a plurality of blocks; performing a comparison operation based on the plurality of blocks and the enrolled partial images corresponding to partial images of an enrolled fingerprint; and recognizing the fingerprint of the first user based on a result of the comparison operation.