Abstract
This article presents an extended author's version based on our previous work, where we introduced the Multiple Overlapping Tiles (MOT) method for palm vein image enhancement. To better reflect the specific operations involved, we rename MOT to ILACS-LGOT (Intensity-Limited Adaptive Contrast Stretching with Layered Gaussian-weighted Overlapping Tiles). This revised terminology more accurately represents the method's approach to contrast enhancement and blocky effect mitigation. Additionally, this article provides a more detailed analysis, including expanded evaluations, graphical representations, and sample-based comparisons, demonstrating the effectiveness of ILACS-LGOT over existing methods.
Introduction
Palm-vein recognition is a highly secure biometric method because vascular patterns are unique, stable over time, and hidden beneath the skin. However, capturing these patterns requires Near-Infrared (NIR) light, which is scattered by skin tissue. This results in blurry, low-contrast images plagued by sensor noise.
Standard enhancement techniques often fall short. Global methods fail to capture fine local details, while local methods like CLAHE often produce "blocky effects"—visible, sharp edges between the small regions (tiles) the image was divided into for processing. These artifacts can be mistaken for actual features by computer vision algorithms, leading to false matches and system failures.
Methodology
The proposed ILACS-LGOT framework addresses these issues through a two-part process designed to sharpen images while eliminating artifacts.
1. Intensity-Limited Adaptive Contrast Stretching (ILACS)
Instead of a standard linear stretch, ILACS uses a modified formula incorporating a floor function. This acts as an automatic clipping mechanism that limits contrast stretching based on local pixel intensities, preventing the system from over-amplifying noise in uniform areas.
2. Layered Gaussian-weighted Overlapping Tiles (LGOT)
To eliminate blocky artifacts, the method employs four overlapping layers of tiles:
- Layer B: The base grid.
- Layers H & V: Tiles centered on the horizontal and vertical edges of the base grid.
- Layer C: Tiles centered on the common vertices (corners) of the base grid.
These layers are blended using a 2D isotropic Gaussian mask acting as an alpha channel. This ensures that the center of each tile has the most influence, while the edges—where artifacts usually occur—fade out smoothly into the overlapping tiles of other layers.
Experimental Results
The researchers evaluated their method using the CASIA Multi-Spectral palm-vein dataset, specifically the 850nm NIR subset.
Contrast Enhancement Analysis
The ILACS component effectively redistributes intensity values to improve local contrast without introducing excessive noise, as evidenced by the broadened intensity distribution in processed tiles.
Recognition Performance
Effectiveness was measured using the Equal Error Rate (EER). ILACS-LGOT achieved an EER as low as 0.4% when paired with RootSIFT features, significantly outperforming established methods like Gabor filters and Local Binary Patterns (LBP). While CLAHE often produced grid-like patterns that interfered with feature detection, ILACS-LGOT produced clean, sharp images suitable for high-security biometric applications.
Conclusion
ILACS-LGOT provides a robust solution for palm-vein image enhancement. By using four layers of tiles, every "edge" in one layer is covered by the "center" of a tile in another layer, effectively mitigating blocky artifacts. Future work will focus on reducing noise sensitivity and refining the Gaussian weighting at the outer edges of the image to prevent peripheral blurring.
Summary updated