U.S. flag

An official website of the United States government, Department of Justice.

NCJRS Virtual Library

The Virtual Library houses over 235,000 criminal justice resources, including all known OJP works.
Click here to search the NCJRS Virtual Library

Improving Verification Accuracy by Synthesis of Locally Enhanced Biometric Images and Deformable Model

NCJ Number
241128
Journal
Signal Processing Volume: 87 Issue: 11 Dated: 2007 Pages: 2746-2764
Author(s)
Richa Singh; Mayank Vatsa; Afzel Noore
Date Published
2007
Length
19 pages
Annotation

This article proposes a two-stage preprocessing framework that increases the verification performance of image-based biometric systems through image enhancement and deformation techniques.

Abstract

In biometrics, quality refers to the intrinsic physical data content, which pertains to the accuracy with which physical characteristics are represented in a given biometric data. The performance of a biometric system depends on the quality of images collected as either a reference or a live sample. This article proposes a SVM-based algorithm that selects good quality local regions from different globally enhanced images and synergistically combines them to produce a high-quality, feature-rich image. The algorithm can be used to remove multiple irregularities present locally in the image without affecting the good quality regions. The authors also propose the phase congruency-based deformation correction algorithm that deforms the query image with respect to the reference image in order to minimize any variations between the two images. The proposed framework can be applied with any of the recognition algorithms in order to improve the verification accuracy. Validation of the framework involved the selection of face and iris images as the two case studies. The preprocessing framework improved the verification performance of face and iris images by 7.6 percent and 1.6 percent, respectively. 4 tables, 20 figures, and 52 references