الفهرس | Only 14 pages are availabe for public view |
Abstract Iris has been widely recognized as one of the strongest biometrics attributed to the high system performance of iris recognition systems. However, templates in conventional iris recognition systems are unprotected and highly vulnerable to numerous security and privacy attacks. A number of iris template protection schemes have been proposed, but at the expense of substantially decreased system performance from the security perspective. In this dissertation, we introduce new cancelable iris template protection schemes. Instead of using original iris features, masked versions of these features are generated for increasing the iris recognition system privacy. The proposed work will be generally divided into five main schemes. These proposed schemes strike the balance between system performance and privacy/security protection. Moreover, evaluation metrics are used for the performance of the proposed schemes. The dissertation objectives will be summarized separately in the following paragraphs. Firstly, we present a random projection scheme for cancelable iris recognition. Instead of using original iris features, masked versions of the features are generated through the random projection in order to increase the security of the iris recognition system. The proposed framework for iris recognition includes iris localization, sector selection of the iris to avoid eyelids and eyelashes effects, normalization, segmentation of normalized iris region into halves, selection of the upper half for further reduction of eyelids and eyelashes effects, feature extraction with Gabor filter, and finally random projection. This framework guarantees exclusion of eyelids and eyelashes effects, and masking of the original Gabor features to increase the level of security. Matching is performed with the Hamming Distance (HD) metric. The proposed framework achieves promising recognition rates and a leading Equal Error Rate (EER). Secondly, a chaos-based cancelable biometric scheme for iris recognition is proposed. The chaotic map encryption is used for generating cancelable IrisCodes for increasing system privacy. A modification of the Logistic map is included to increase the key space, and hence the privacy is enhanced. The encryption key depends on the input image. Hence, the resultant encrypted feature vector is sensitive to the key. Thus, the encrypted feature vector is robust as the key space is large. This proposed scheme achieves a high accuracy and a good EER upon using the modified Logistic map on the CASIA-IrisV3 dataset. Thirdly, we implement the optical Double Random Phase Encoding (DRPE) algorithm in cancelable face and iris recognition systems. In the proposed cancelable face recognition scheme, the Scale Invariant Feature Transform (SIFT) is used for feature extraction from the face images. The extracted feature map is encrypted with the DRPE algorithm. On the other hand, the proposed cancelable iris recognition system depends on the utilization of two iris images for the same person. Features are extracted from both images. The features extracted from one of the iris images are encrypted with the DRPE algorithm provided that the second phase mask used in the DRPE algorithm is generated from the other iris image features. This trend guarantees some sort of feature fusion between the two iris images into a single cancelable iris code and increases privacy of users. Simulation results show a good performance of the two proposed cancelable biometric schemes even in the presence of noise, especially with the proposed cancelable face recognition scheme. Fourthly, we present a new technique for cancelable iris recognition using Comb filtering approach. In this technique, all enrollment patterns are masked using a transformation function, and the invertibility process for obtaining the original data should not be possible. Experimental results are con are have been calculated for different values of Comb filter orders and compared with the unprotected IrisCode results. Hamming distance and Receiver Operating characteristic (ROC) distributions are estimated for different Comb filter orders to check the system robustness and stability. The experimental results achieve a significant gain for both privacy and performance proving the superiority of the proposed scheme. Also, the proposed scheme achieves a high accuracy and a promising EER.Lastly, we present a novel CNN model that successfully classifies different scenarios of cancelable biometric traits using a bio-convolving method applied on both image and feature levels. On the contrary of most conventional secure recognition systems, the proposed scheme for cancelable biometric recognition maintains high accuracy results, while maintaining cancelability. The performance metrics are evaluated for different traits with different datasets; LFW, FERET, IITD, and CASIAIrisV3. The experimental results are demonstrated for each database and are compared with those of the state-of-the-art methods applied to the same databases. The results show the robustness and effectiveness of the proposed scheme. Moreover, it shows high recognition rates for all databases. |