![]() | Only 14 pages are availabe for public view |
Abstract Recently, an outstanding performance was achieved by biometric systems based on adults’ characteristics like iris, fingerprint, face, palm print and footprint. In the last years, there were a lot of applications specifically for adults. However, young children recognition has a lot of issues and is not recorded in every age group. To ensure that young children have their rights, they must be identified to be recognized in front of the law, help reunite missing children with their parents. Also, it puts a stop to changing children in any medical institution, recognizing lost young children and civilian identity applications.This thesis presents a modified multimodal biometric recognition for toddlers and pre-school children (M²BRTPC) technique based on the characteristics and temporal stability of biometric recognition accuracy for children. The technique uses iris and fingerprint images from the children multimodal biometric database (CMBD) of over than 100 children (aged 1 and half to 4 years old) ignoring face modality based on low recognition accuracy results. Also, a new fingerprint extraction tool (FPET) has been presented to enhance the accuracy of the fingerprint modality. The M²BRTPC approach consists of two stages: enrollment and testing modules. The enrollment module consists of four steps which are; (I) image acquisition, (II) image preprocessing, (III) feature extraction system, and (IV) template generate that stores the extracted features in a new database. The testing module consists of five steps; the same previous four steps in addition to the fifth one (V) template matching system, after generating the template it searches within the created database to find a match. Experimental results have shown the effectiveness of the proposed approach techniques. Different experiments have been performed to compare the performance of the new approach including single modal experiments and multimodal fusion experiments. The experimental results demonstrate an enhancement in the M²BRTPC approach in single and multimodal fusion experiments compared to researches that uses the same database. |