Maha Sharkas

Eigenfaces vs. fisherfaces vs. ICA for face recognition; a comparative study

Face recognition issue gained more interest recently due to its various applications and the demand of high security. Some researches with contradicting results were published concerning this issue. This paper compared three popular face recognition projection methods: (eigenfaces), (fisherfaces), and ICA. We also applied some data transformations: (discrete wavelet and cosine transforms) preceding methods to see their effect. Most researches based their results on the FERET database. AR and AT & T databases were used here to see if the same results apply. We also compared the results of two sets of experiments with the second set using half the training images used in the first to observe if the results may change. Overall conclusion is it canpsilat be stated that specific algorithm outperforms others, though ICA and eigenfaces respectively showed better results than fisherfaces for both experiments sets and both databases. Preceding algorithms with transformations yield better results for some algorithms.

The Contourlet Transform with the Principal Component Analysis for Palmprint Recognition

In this paper two techniques for palmprint recognition are suggested and compared. Palmprint include principal lines, wrinkles and ridges which contain very important features essential for recognition. The Contourlet Transform (CT) is a multiresolution and multidirection transform which can be effective in capturing the palm features. The first technique extracts the edges from the palm images and then performs the CT or the Discrete Wavelet Transform (DWT) on the edge extracted images. The sub-band images are divided into M*M non-overlapping blocks. The energy of each block is calculated and normalized to form a feature vector. The second technique employs the principal component analysis PCA where the approximation images are input to it for dimensionality reduction and to produce the eigen palms. Features extracted from both techniques are tested and compared where it was found that the best achieved recognition rate is about 94% when combining the results of both techniques using the CT.This paper analyzes the problems in traditional manufacturing and assembly, puts forward the Estimation Indication for Cost of Quality Losing in an optimization assembly which will be based on RFID. Though this assembly is a kind of selective one, the whole production will be according to the law of interchangeability. The quality of assembly will be raised, but the cost of manufacturing will be decreased or not increased.

Detection of Microcalcifications in Mammograms Using Support Vector Machine

For years cancer has been one of the biggest threats to human life, it is expected to become the leading cause of death over the next few decades. Early detection of breast cancer can play an important role in reducing the associated morbidity and mortality rates. Clusters of micro calcifications (MC) in the mammograms are an important early sign of breast cancer. Mammography is currently the most sensitive method to detect early breast cancer. Manual readings of mammograms may result in misdiagnosis due to human errors caused by visual fatigue. Computer aided detection systems (CAD) serve as a second opinion for radiologists. A new CAD system for the detection of MCs in mammograms is proposed. The discrete wavelet transforms (DWT), the contour let transform, and the principal component analysis (PCA) are used for feature extraction, while the support vector machine (SVM) is used for classification. The best classification rate was achieved using the DWT features. The system classifies normal and tumor tissues in addition to benign and malignant tumors. The classification rate was 100%.

An Energy-Based Dual Image Watermarking Technique with Application to Color Images

A novel adaptive dual image watermarking technique is suggested and tested. The technique embeds a PN sequence which is the primary watermark into an image (a secondary watermark) and the resulting image is then embedded in the host image. The technique is implemented in the wavelet domain and the embedding factor alpha is first chosen arbitrary so as to improve the invisibility and robustness and then chosen adaptively depending on the energy content of the image to be watermarked in order to improve the performance. The technique is implemented on several gray scale images and then on several color images. The best achieved peak signal to noise ratio (PSNR) in case of gray scale images reached 68.8432 db whereas in color images it was 54.8750 db.

A SPIHT/OFDM with Diversity Technique for Efficient Image Transmission Over Fading Channels

In this paper, we propose a combined SPIHT/OFDM coding scheme for transmission of image/video over wireless fading channels. We take a close look at the performance of MRC diversity reception of OFDM signals in multipath slowly fading Nakagami-m channels. The proposed system implements a rate control algorithm as the image coding scheme and an adaptive modulation OFDM system as the communication coding scheme. We evaluate the system over uncorrelated Nakagami-m fading channel. Simulation results for peak signal-to-noise ratio (PSNR) of the received image at different rates are presented which demonstrate the contribution of our algorithm over this challenging channel.

An enhanced WiFi indoor localization system based on machine learning

The Global Navigation Satellite Systems (GNSS) suffer from accuracy deterioration and outages in dense urban canyons and are almost unavailable for indoor environments. Nowadays, developing indoor positioning systems has become an attractive research topic due to the increasing demands on ubiquitous positioning. WiFi technology has been studied for many years to provide indoor positioning services. The WiFi indoor localization systems based on machine learning approach are widely used in the literature. These systems attempt to find the perfect match between the user fingerprint and pre-defined set of grid points on the radio map. However, Fingerprints are duplicated from available Access Points (APs) and interference, which increase number of matched patterns with the users fingerprint. In this research, the Principle Component Analysis (PCA) is utilized to improve the performance and to reduce the computation cost of the WiFi indoor localization systems based on machine learning approach. All proposed methods were developed and physically realized on Android-based smart phone using the IEEE 802.11 WLANs. The experimental setup was conducted in a real indoor environment in both static and dynamic modes. The performance of the proposed method was tested using K-Nearest Neighbors, Decision Tree, Random Forest and Support Vector Machine classifiers. The results show that the performance of the proposed method outperforms other indoor localization reported in the literature. The computation time was reduced by 70% when using Random Forest classifier in the static mode and by 33% when using KNN in the dynamic mode.

MBLBP face detection with multi-exit Asymmetric Boosting

Face detection plays an important role in many applications such as video surveillance, face recognition, face image database management etc. This paper presents a new technique which reduces the learning and detection time using the multi block local binary pattern (MBLBP) with Multi-exit Asymmetric Boosting. In this technique, the selected features are reduced by around 1/20 of Haar-like method so the learning time is also reduced by about 1/20. The detection time is also reduced by more than 1/4 of Haar-like detector. Multi-exit Asymmetric Boosting reduces features by about 1/5 of the cascade method so the learning and detection time is also reduced.

Identification of transmission line faults using wavelet analysis

Transmission line fault identification requires fast and accurate analysis. The tripping action depends mainly on the voltage and current waveforms during the fault. Wavelet analysis which is a mathematical tool for signal analysis is used to detect the type of fault occurring on the transmission line. According to this analysis tripping decision is processed. The PSCAD which is an EMTP program with graphical support is used for the simulation of the power system under consideration. Discrete wavelet transform (DWT) is used for the analysis of the current waveform during the fault. Successful testing of the proposed technique proves its validity for near perfect detection of the fault type.

Priority level mutualism for emergency vehicle using game theory

Congestion growth problem is a daily life experience all over the world, especially in large cities. The motion of Emergency Vehicles (EVs) like Ambulances, police, and firefighting vehicles are highly affected by traffic Congestion, by which rescue time may cost a life. The Priority Level Mutualism for Emergency Vehicle (PLMEV) algorithm aims to reduce the waiting time by giving higher priority for the EVs at the intersection. According to the vehicles priorities, The Game Theory will control the traffic at the intersection. The PLMEV classifies EVs according to their emergency levels. The algorithm has a highly scalable property with a different number of vehicles. Also, has adaptive property to different vehicle type distribution and work with multiple EVs. Simulation evaluates the PLMEVs performance by comparing it with the Fixed Traffic Control System. The Simulations results proved that The PLMEV reduced the waiting time for EVs to 42.7% on average, on the other hand, the normal vehicles time has been reduced to 15.1% on average, while the maximum vehicle waiting time is reduced to 22.3% on average. In case there is no EVs at the intersection, the PLMEV works in optimising the traffic flow.

Adaptive forward-backward orthogonal matching pursuit for compressed sensing

In this paper we propose a novel iterative greedy algorithm for solving under-determined linear system of equations y = Ax when the solution vector x is known a priori to be sparse. The proposed algorithm falls into the general category of two stage thresholding (TST) algorithms. The proposed algorithm follows an iterative procedure to estimate the support of the sparse solution vector in a dynamic way. Therefore, it has the capability of correcting any indices of the estimated support that were erroneously incorporated in early stages. The proposed algorithm depends on a parameter a called the forward step-size. In this paper we propose an approach for computing the value of a adaptively in each iteration. Following this approach, the simulation results show that the proposed algorithm outperforms state of the art algorithms used for solving the same problem.