Enhance Criminal Investigation by Proposed Fingerprint Recognition System

Soukaena H. Hashem; Abeer T. Maolod; Anmar A. Mohammad

Soukaena H. Hashem Computer Sciences Department, University of Technology, Baghdad/Iraq.
Abeer T. Maolod Computer Sciences Department, University of Technology, Baghdad/Iraq.
Anmar A. Mohammad Computer Sciences Department, University of Technology, Baghdad/Iraq.

Keywords: Crime Analysis, Criminal, fingerprint, minutiae, genetic algorithm, feature extraction, FRR, FAR.

Abstract

Law enforcement officers and forensic specialists spend hours thinking about how fingerprints solve crimes, and trying to find, collect, record and compare these unique identifiers that can connect a specific person to a specific crime. These individuals understand that a basic human feature that most people take for granted, can be one of the most effective tools in crime solving. This research exploits our previous work to be applicable in criminal investigation field. The present study aims to solve the advance crime by strength fingerprint’s criminal investigation to control the alterations happen intentionally to criminals’ fingerprint. That done by suggest strategy introduce an optimal fingerprint image feature’s vector to the person and then considers it to be stored in database for future matching. Selecting optimal fingerprint feature’s vector strategy deal with considering 10 fingerprints for each criminal person (take the fingerprint in different time and different circumstance of criminal such as finger is dirty, wet, trembling, etc.). Proposal begun with apply a proposed enrollment on all 10 fingerprint for each criminal, the enrollment include the following consequence steps; begin with preprocessing step for each of 10 images including enhancement, then two level of feature extraction (first level to extract arches, whorls, and loops, where second level extract minutiae), after that applying proposed Genetic Algorithm to select optimal fingerprint, master fingerprint, which in our point of view present the most universal image which include more detailed features to recognition. Master fingerprint will be feature’s vector which stored in database. Then apply the proposed matching by testing fingerprints with these stored in database. While, measuring of criminal fingerprint investigation performance by calculating False Reject Rate (FRR)and False Accept Rate (FAR) for the traditional system and the proposed in criminal detection field. The obtained results encourage to publish this work.