3D Finger Vein Recognition via Multi-View Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vein recognition technologies face challenges in achieving robust recognition across various finger poses due to limited feature information and poor image quality, especially when using two-dimensional imaging methods, which are sensitive to rotation and displacement deviations.
Innovation Solution
A three-dimensional finger vein recognition method using three cameras to capture images from evenly spaced angles, constructing a three-dimensional finger model, mapping two-dimensional image textures, and performing feature extraction and matching on the resulting three-dimensional image to enhance recognition robustness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a monocular camera is used to perform two-dimensional imaging on a single side of a finger, then the device complexity is reduced, but the quantity of vein feature information is limited and recognition robustness deteriorates
Solution Approach 1:
The patent transitions from two-dimensional single-view imaging to three-dimensional multi-view imaging by adding spatial dimensions. Multiple cameras capture vein images from different angles (front, side, top views), and these 2D images are integrated to reconstruct 3D finger vein models, thereby enriching the vein feature information without excessive complexity increase
Solution Approach 2:
The imaging process is segmented into multiple independent views (front view, side view, top view), each captured by separate cameras. This segmentation allows comprehensive coverage of finger vein features from different perspectives, and the segmented images are subsequently integrated to form complete 3D vein models
2Device complexity
If two-dimensional imaging is used, then the device complexity is reduced, but the recognition precision deteriorates under finger rotation and displacement
Solution Approach 1:
By adding the third dimension (depth) through multi-view 3D imaging, the system creates rotation and displacement invariant features. The 3D vein models preserve spatial relationships that remain consistent regardless of finger pose changes, thereby improving recognition precision under varying conditions
Solution Approach 2:
The patent changes the imaging parameters from 2D coordinates to 3D spatial coordinates, and from single-view perspective to multi-view perspectives. This parameter transformation enables the extraction of pose-invariant features that maintain recognition precision across different finger positions and orientations
3Loss of information
If three-dimensional multi-view imaging is implemented, then the quantity of vein feature information increases, but the device complexity increases
Solution Approach 1:
The complex imaging task is segmented into multiple independent but coordinated camera systems, each capturing specific views (front, side, top). This segmentation allows modular design and independent optimization of each camera unit while achieving comprehensive 3D coverage when integrated
Solution Approach 2:
The multi-view imaging system serves multiple functions simultaneously: capturing vein patterns from different angles, providing depth information, enabling 3D reconstruction, and creating rotation-invariant features. This multi-functionality justifies the increased device complexity by delivering comprehensive vein information
4Measurement precision
If three-dimensional reconstruction is performed using multi-view geometry, then the measurement precision improves, but the difficulty of detecting and measuring increases due to difficulty in finding matched feature points
Solution Approach 1:
The patent performs preliminary processing of vein images before 3D reconstruction, including image enhancement, vein segmentation, and feature point detection in each 2D view. This preliminary action prepares the data in advance, making the subsequent 3D reconstruction and feature matching more efficient and accurate
Data Source
AI summary
A three-dimensional finger vein recognition method and system, comprising the following steps: three cameras taking finger vein images from three angles to obtain three images; constructing a three-dimensional finger model according to finger contour lines; mapping two-dimensional image textures photographed by the three cameras into the three-dimensional finger model, respectively performing different processes on an overlapping region and a non-overlapping region; obtaining a three-dimensional finger vein image; and finally, performing feature extraction and matching on the three-dimensional finger vein image, to complete recognition. The method can acquire a better finger vein recognition effect, and has a higher robustness for a plurality of postures, such as finger rotation and inclination.