“Where uniqueness speaks...”

Personal

More about me

 

Click HERE for the PDF version of my resume (updated!!!)

My publications can be downloaded HERE

 

 

EDUCATION

PROFESSIONAL EXPERIENCES

RESEARCH EXPERIENCES

*3D Touchless Fingerprints

Develop a non-parametric unwrapping technique for virtual rolling of 3D touchless fingerprints and demonstrate their compatibility with legacy rolled fingerprints. The proposed virtual rolling process also shows compatible deformation under different rolling directions.

* Extended Fingerprint Feature Set

Design and develop algorithms to extract extended features such as pores and ridge contours for 1000ppi fingerprint matching. The iterative control point (ICP) algorithm is used for matching extended features. An automated hierarchical system that utilizes multi-level features for is also employed to improve matching accuracy and efficiency.

*Image Quality Assessment for Biometric Systems

A multi-biometric quality checker was developed to examine the quality of fingerprint and iris images. The fingerprint checker evaluates both local and global quality of a fingerprint image using FFT and coherence measure, respectively. The iris checker measures the localized iris features using a 2D wavelet-based algorithm. A quality-based multi-algorithm fingerprint system is also implemented to demonstrate the effects of image quality on fution performance.

*Fingerprint Deformation Models

Fingerprint deformation estimation for the purpose of removing non-linear deformations prior to the matching stage. A finger-specific deformation model is constructed using the Thin-Plate Splines and an optimal template selection strategy is established based on the bending energy.

*Fingerprint Spoof Detection

Non-linear deformation is a natural choice for spoof detection since human skin has special elastic property, compared to artifical materials, such as gelatin, silicon, etc. Based on the differences of the non-linear deformation estimated between live vs. live pairs and live vs. fake pairs, we are able to detect fake fingerprints at a high classification rate.

*Fingerprint Registration

Establish minutiae correspondences using a voting scheme and utilize alignment of orientation fields and singular points to improve the robustness of our current matcher.

*Slap Fingerprint Segmentation and Sequence Checking

Develop algorithms to segment slap fingerprint images into individual fingerprint images and conduct a sequence checking procedure to match rolled fingerprints against segmented slap fingerprints for correct sequencing assurance.

*Iris Image Segmentation

Iris region is segmented from the eye image for the purpose of iris coding. This involved image enhancement, wavelets and Hough transform.

PUBLICATIONS