Abstract:
Multimodal biometric systems have continued to demonstrate ability to fill the gap created by the numerous limitations of the unimodal biometric systems. Such limitations include degrading performance with noisy data, susceptible to spoof attacks, non-universality, non-individuality, intra-class similarities and so on. Notwithstanding, the accuracy of multimodal biometric system is highly dependent on the adequacy of the applied fusion technique. Fusion at sample, template, matching and ranking levels have all proved reasonable contributions to the performance of the multi-modal systems. In this research work, a combination of Principal Component Analysis (PCA) and Stationary Wavelet Transform (SWT) based model for the fusion of biometric images is proposed. The model requires the preprocessing stage of Gaussian filtering, histogram equalization, cropping and normalization as well as PCA-based feature extraction. The decomposition and fusion of the images (based on the extracted features) were based on SWT. The experimental study of the model with standard face and ear images revealed that the model is very suitable for obtaining fused images with high quality. The obtained Peak Signal to Noise Ratio (PSNR), Mean Square Error (MSE) and Standard Deviation (SD) values established the superiority of the proposed model over some related ones.
