Optimized multimodal biometric system based fusion technique for human identification
Bulletin of Electrical Engineering and Informatics
This paper presents three novelty aspects in developing biometric system-based face recognition software for human identification applications. First, the computations cost is greatly reduced by eliminating the feature extraction phase and considering only the detected face features from the phase congruency. Secondly, a motivation towards applying a new technique, named mean-based training (MBT) is applied urgently to overcome the matching delay caused by the long feature vector. The last
... ty aspect is utilizing the one-to-one mapping relationship for fusing the edge-to-angle unimodal classification results into a multimodal system using the logical-OR rule. Despite some dataset difficulties like unconstrained facial images (UFI) which includes varying illuminations, expressions, occlusions, and poses, the multimodal system has highly improved the accuracy rate and achieved a promising recognition result, where the decision fusion is classified correctly (84, 92, and 72%) with only one training vector per MBT in contrast to (80, 62, and 68%) with five training vectors for normal matching. These results are measured by Eucledian, Manhattan, and Cosine distance measure respectively. 2412 RELATED WORKS P. Kovesi was the most interested in the PC field since 1999, a calculation for 1-D signal was extended to 2-D images using wavelet high-pass filters to obtain image information at different scales where a universal threshold value over wide class of images was applied  . The advantages of using PC over Canny edge detectors was explained and approved using standard images  . PC-based face recognition technique for improving the recognition rates of the faces that are affected due to varying illuminations, partial occlusions, and varying expressions was proposed and implemented  . A descriptor based on the PC concept, called histogram of oriented phase (HOP) was applied to depict and represent the human objects more efficiently than the gradient based approach  . A modular kernel eigen spaces approach was implemented on the PC images to localize nonlinear feature selection procedure to help overcoming the bottlenecks of illumination variations, partial occlusions, and expression variations  . A 2-D multi-scale phase congruency (2D-MSPC) software for detecting and evaluation of image features could appropriately tune many parameters for optimal image features detection, these parameters are optimized for maximum and minimum moments . A modified algorithm of PC to locate image features using Hilbert transform was implemented and the local energy was obtained by convoluting original image with two operators of removing direct current (DC) component over current window and 2-D Hilbert transform respectively. The local energy was divided with the sum of Fourier amplitude of current window to retrieve the value of PC  . A quality matric for evaluation of different video fusion methods using set of predefined 3D Gabor filters was employed to compute the spatial temporal PC for input and fused videos  . The local binary pattern (LBP) on the ultra sound (US) medical images after computing their PCs was used to improve the reliable feature point localization  . The multi-focus image fusion applied new fusion rule and complex Gabor wavelet to obtain the benefits of PC sharpness in finding new focus measure  . Two different multi-spectral image fusion rules for nonsubsampled contourlet transform (NSCT) was introduced with: PC, principle component analysis (PCA), directive contrast, and entropy for developing integrity model  . A fusion of the bio-medical images and guided filter was implemented by decomposing the bio-medical images into two sub-bands using the non-subsampled contourlet transform. The guided filter was very effective in extracted the high frequency details for obtaining the final fused image in spatial domain  . The edges in the liver image was enhanced by improving the segmentation accuracy. The sum modified Laplacian (SML), contrast features measure, and non-subsampled shearlet transform (NSST) were integrated to reconstruct a model for better fusion of edge information  . Pearson correlation coefficient and Spearman rank order correlation coefficient  were used to evaluate the performance of the proposed metric. The proposal developed a novel image quality and metric as a set of ten extracted features from each distorted image, then relevance vector machine algorithm (RVM) was used to learn the mapping between the human opinion scores and combined features. A novel descriptor using multioriented PC and magnitude information was presented to show the advantages of the histograms of oriented magnitude (HOM) and Histograms of oriented phase congruency (HPC) combination, also it confirmed that the histograms of oriented magnitude and phase congruency (HOMPC) was so superior to the local feature in the state-of-the-art descriptors. Three datasets were composed of multi-sensor remote sensing images obtained from satellite platforms, unmanned aerial vehicle, and unmanned ground vehicle  . An immune model of PC against illumination and contrast variation was developed. An obtained result from variety of remote sensing images showed an excellent performance against the local invariant and robust of the model to both geometric and radiometric changes  . The negatively impact was avoided using the same fusion rule on different scales, also a Gaussian filter with multi-scale decomposition (MSD) of total variation (TV) and PC were proposed for designing the modal  .