Optimal feature reduction for biometric authentication using intelligent computing techniques

The Intelligent Computing area such as Automatic Biometric authentication is an emerging and high priority research work where the researchers invent several biometric applications which result in the revolutionary development in the recent era. In this approach, a novel algorithm is known as Modified AntLion Optimization (MALO) with Multi Kernel Support Vector Machine (MKSVM) was used to classify and recognize the fingerprint, and retina image efficiently. In the early stage of this research, the pre-processing of the biometric images was done for contrast enhancement and it was implemented by histogram equalization technique. Next, features were extracted by Gray Level Co-occurrence Matrix (GLCM), minutiae, Gray Level Run Length Matrix (GLRLM), and Autocorrelation methods. Then the features extracted were reduced by Probabilistic Principal Component Analysis (PPCA) method. Then the feature selection method was employed and the optimal features were attained by applying the Modified AntLion Optimization (MALO) technique. Finally, the machine learning classification technique was executed for categorizing biometric recognition. Here, the machine learning classification technique named Multi Kernel Support Vector Machine (MKSVM) has been used. The performance of the proposed algorithm was analyzed in terms of accuracy, sensitivity, and specificity. Results indicate that the Multi Kernel Support Vector Machine (MKSVM) yields the best accuracy of 91.60% and 90.30% for fingerprint and retina image recognition respectively, yields the sensitivity of 84.70% and 89.41% for fingerprint and retina image recognition, respectively, yields the specificity of 91.30% and 92.70% for fingerprint and retina image recognition respectively.