A neoteric approach for ear biometrics using multilinear PCA

In the current era of Information technology, almost every device and application requires Internet connectivity. Security and access to data has become a major area of concern. Since each and every algorithm which use different transformation method are better at some place and same time has some limitation. So enhanced approach can implement the hybrid method to achieve the more accuracy and inherit the best property of the transforms. We use MPCA (Multi linear PCA) algorithm to generate feature vector and after applying multiple transforms to the image we use a Log Gabor filter which will help in accurate and efficient feature extraction and help the MPCA to generate vectors for matching similarity. The significant contributions of this project are design and implementation of fusion of ear and soft biometric for recognition of people and preparation of ear and soft biometric database. Fusion of ear and soft-biometrics result in a development over the primary biometric system. In the proposed system ear is a unimodal biometric authentication system which extract unique feature and give recognition accuracy of 82.5% while combining of unimodal (ear) with soft features(height and gender) is a unimodal soft biometric which give recognition rate of 92.3%. The experimented tests demonstrated that there is an improvement to recognize genuine users as high True Accept Ratio (TAR) and less False Accept Ratio (FAR). Proposed methodology shows perfect index of ear with soft biometric is 92.3%,which is 10% more than individual primary (ear) biometric system. While capturing images of each subject: illumination variation, pose variation, distance variation, date-variation and occlusion variation Size of database will increase. To make the enrolment process automatic there is a need to construct a model of variation.