UPM Institutional Repository

Efficient hardware design for palm-dorsa vein image enhancement


Jusoh@Yusoff, Suhaimi Bahisham (2018) Efficient hardware design for palm-dorsa vein image enhancement. Doctoral thesis, Universiti Putra Malaysia.


Vein biometric system uses the pattern of veins in the human body as a unique identification. In this thesis, vein at palm-dorsa has been used as biometric modal. Nowadays, many researchers have developed numerous algorithms to enhance vein image for biometric purposes with most researchers focusing only on the use of software such as MATLAB or C language. The main problem with software implementation is that it cannot achieve the high computational speed required in specific application such as real time application that requires instant authentication. Hardware implementation can provide a solution to achieve real time performance. Recent hardware design implementations of vein image enhancement are not fast enough due to the inefficient hardware design architecture. The aim of this thesis is to develop hardware design for palm-dorsa vein image enhancement algorithm. The algorithm is designed in hardware using hardware description language for hardware realization. ModelSim-Altera has been used as hardware simulation platform. First, the vein image is applied with resample technique to remove the noise. Then, segmentation technique consisting of Difference of Gaussian and threshold are used to segment the veins. After that, median filter is used to remove noise introduced from the image segmentation. Finally, thinning technique is applied to get single line vein. For hardware design of resample technique, parallel pipeline hardware has been developed to improve processing time and high throughput. It is designed to perform bicubic computation in parallel pipeline hardware to accommodate fast processing and high throughput with less hardware resources. For the interpolation process, instead of using multipliers, shifters are used to reduce hardware resources. For hardware design of Difference of Gaussian, one dimensional Gaussian technique that operated concurrently for first Gaussian filter and second Gaussian filter are implemented. For threshold, a simple comparator has been used to design threshold in hardware. For hardware design of median filter, the improved moving windows hardware architecture has been developed. In the improved moving windows, instead of calculating all the pixels in the window, only certain pixels are calculated. The hardware design architecture of resample, segmentation and median filter also feature padding capability by reading the same address of memory as before for the padding pixel. For hardware design of thinning, parallel pipelined with Concurrent Condition Check Unit hardware architecture has been developed to enable the parallelism of the thinning algorithm. It contains modules that executed the thinning algorithm function simultaneously in hardware to speed up the process. Finally, hardware design architecture of vein image enhancement algorithm has been proposed by integrating the hardware designs of resample, segmentation, median filter and thinning. The findings show that the proposed hardware design has the percentage of correct is about 98%. The proposed hardware design has the execution time of 8.5ms to 12.4ms depending on the thinning iterations. This work contributed in efficient hardware design architecture for palm-dorsa vein image enhancement for biometric purpose.

Download File

FK 2018 68 IR.pdf

Download (2MB) | Preview

Additional Metadata

Item Type: Thesis (Doctoral)
Subject: Biometric identification - Case studies
Subject: Image analysis
Subject: Blood-vessels - Imaging - Case studies
Call Number: FK 2018 68
Chairman Supervisor: Abd. Rahman bin Ramli, PhD
Divisions: Faculty of Engineering
Depositing User: Ms. Nur Faseha Mohd Kadim
Date Deposited: 27 Nov 2019 07:49
Last Modified: 29 Nov 2019 01:43
URI: http://psasir.upm.edu.my/id/eprint/76055
Statistic Details: View Download Statistic

Actions (login required)

View Item View Item