Make biometric-based person identification system applicable

• Main Author: Gan, David Jian Wen
• Other Authors: Li Fang
• Format: Final Year Project
• Language: English
• Published: 2012
• Subjects: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Pattern recognition; DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
• Online Access: http://hdl.handle.net/10356/49124
• Institution: Nanyang Technological University

Biometric is the study of measurable biological characteristics of identifying a person using their physical or behavioral characteristics. There are several types of biometric identification schemes such as analysis of individual unique fingerprint and palm print, analysis of capillary vessels of retina to analyzing the way a person signs his signature. In this project, the existing Biometrics Identification System which performs the palm print image identification will be analyzed and different experiments will be concluded to explore the sensitivity, effectiveness and adaptability of the image preprocessing and image matching algorithm. Besides using palm print image as the biometric identification measure, vein pattern images will also be used as well. Experiments using 100 palm print images results in a high matching accuracy of 98%. This result gives a high confidence level about the existing palm print preprocessing and identification algorithms and different experiments can be further performed using different quality images, database size and different biometry measure such as vein print image. Another experiment using a database of 200 palm print images which are collected using mobile phone show a relatively good matching accuracy result of 83.5%. This experiment results show a good accuracy rate for palm print image database and the mobile phone palm print image database despite the different quality of the palm print images captured by the two databases. Besides using lower quality image, experiments have also been concluded to test the sensitivity of the algorithm when the palm print images have been modified or distorted. The results of the experiment show a high 97% match when palm print images have been skewed to 10 degrees. However, the accuracy of the image matching is not good when there are missing areas or unknown lines added to the image. Besides using the palm print images as a biometric identification measure, vein pattern images biometry will also be used to test the adaptability of the existing algorithms. The results of the identification of the vein pattern images shows an 87.5% accuracy when 200 vein print images have been used. The high accuracy results of the vein pattern images shows that the existing system can be used to perform image matching for both the palm print and vein pattern images. These experiments show that the image preprocessing and image matching algorithms for the Biometric Identification System will be able to perform accurate image matching for both the palm print and vein print. Detailed analysis of the accuracy results for all the experiments for the different databases will be analysed in the later sections of the report. This report also discusses the overview of the Biometric Identification System for the palm print and vein pattern from the preprocessing stage to the image matching stage.