Combining multi-scale composite windows with hierarchical smoothing strategy for fingerprint orientation field...

Area of Science:

• Biometrics
• Computer Vision
• Image Processing

Background:

• Orientation field (OF) estimation is crucial for automatic fingerprint recognition.
• Existing methods struggle with balancing accuracy and noise resistance, and preserving singularity localization, especially in low-quality images.
• Singular point deviation in OF estimation can lead to increased matching errors.

Purpose of the Study:

• To propose a novel method for robust and accurate fingerprint orientation field (OF) computation.
• To address the limitations of existing methods in handling spurious ridge structures and singularity location deviation.
• To improve the reliability of OF estimation for practical fingerprint recognition applications.

Main Methods:

• A weighted multi-scale composite window (WMCM) approach for approximate OF estimation.
• A hierarchical smoothing strategy involving two- and three-orientation-zone filtering for OF refinement.
• Gradient-based methods combined with adaptive scale selection and filtering techniques.

Main Results:

• The proposed method demonstrates superior robustness against singularity localization deviation compared to traditional gradient-based methods.
• Hierarchical smoothing effectively corrects spurious ridge structures while preserving genuine singularity localization.
• The approach shows reliable OF extraction and improved performance on low-quality fingerprint images.

Conclusions:

• A novel gradient-based algorithm offers reliable fingerprint OF estimation and accurate singularity localization.
• The weighted multi-scale composite window adaptively selects block scales, enhancing accuracy and anti-noise capabilities.
• Hierarchical smoothing refines the OF, correcting spurious ridges and avoiding singularity deviation, outperforming existing methods.