Main Article Content
Secure data access model on Mobile cloud data has received a lot of attention recently and is viewed as a promising trait. Biometric based authentication to secure access of the mobile cloud data have undergone fast development towards data protection and preventing malicious data threats. In this work, multimodal biometric based authentication technique as fusion model has been projected for data security. The key characteristics of finger print and face patterns include its uniqueness to each individual, unforgettable, non-intrusive and cannot be taken by an unauthorized person. In related methods, feature selection methods explore intrinsic finger print and face on single feature method, but their performance remains undesirable in terms of computational cost. However the extracted features from the both finger print and face pattern are huge with high redundancy. On employment of fusion concept on feature extraction techniques through weighted average strategies, equal error rate is minimized to obtain the optimum weight. In this paper, we propose a combinational model composed differential evolution technique to enhance the recognition of finger print and face patterns to authenticate the user towards data access. The system has been trained in selecting high relevant features on using the extraction techniques such as principle component analysis and linear discriminant analysis. Feature fusion carried out as concatenation on PCA and LDA technique using discriminant correlation analysis feeds the proposed feature selection models by optimal subset of features. The proposed system uses differential model to determine the less no of optimal features for multimodal authentication. Multimodal authentication performed using Multivariate linear regression on less no of optimal features. The analysis of empirical results shows that proposed system produces the best accuracy reflected by 100 percent accuracy on comparing with existing single model biometric authentication models. Further Proposed model has been evaluated and the results shows remarkable efficiency with existing state of art approaches.
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.