Mohamed R. M. Rizk

Analysis of neural networks for face recognition systems with feature extraction to develop an eye localization based method

This paper provides an analysis of multilayer perceptron backpropagation neural networks (MLP/BP NN), radial basis function neural networks (RBF NN), and multilayer cluster neural networks (MCNN) applications in face recognition. Feature extraction methods involved in the analysis are the discrete wavelet transform (DWT), discrete Radon transform (DRT), discrete cosine transform (DCT) and the principal component analysis (PCA) technique. Algorithms were developed using Matlab and tested on the ORL database. Also, a new proposed 2-stage face recognition system is presented based on eye localization and a windowed face area for recognition.

Electron microscopic and chemical study of the deposits formed on the copper and inert IUCDs

The surface of 25 copper IUCDs and 25 inert IUCDs that had been in utero for less than 6 months or more than a year was examined by the scanning electron microscope and the deposits were extracted from another 100 copper and inert IUCDs that had been in utero for less than 6 months or more than a year and examined by the atomic analyser. The most apparent changes were, respectively, amorphous deposits and cellular debris, crystalline deposits, cracks, duplex coating, distortion and corrosion. Duplex coating was found only on inert IUCDs, but corrosion was found on copper IUCDs. The duration of use of the IUCDs had marked effect on the type of the deposits. Bleeding with IUCDs was associated with a thick layer of amorphous deposits. The contraceptive effect of the IUCDs was found to depend on the reaction of the host to the applied IUCDs rather than the type of device. The surface deposits were found to contain mainly calcium, phosphorus and magnesium.

Efficient contention-based MAC protocol using adaptive fuzzy controlled sliding backoff interval for wireless networks

Efficient medium access control (MAC) protocol should be able to provide high throughput performance and efficient share of the medium. In this paper, a new contention-based MAC protocol based on adaptive fuzzy controlled sliding backoff interval is proposed to maximize the channel throughput and improve the fairness of random access channels. In the proposed protocol, every node that experiences packet collisions increases its sliding backoff interval (SB) range by a forward sliding factor (FSF). In case of successful transmission the node decreases its SB range by a backward sliding factor (BSF). Forward and backward sliding factors are controlled by the channel offered traffic using a fuzzy controller. Furthermore, the operation of the proposed backoff algorithm does not depend on the knowledge of the number of active nodes. A computer simulation is developed using MATLAB to evaluate the performance of the proposed algorithm and compare it with other backoff schemes. Simulation results show that the proposed algorithm significantly outperforms other backoff schemes, such as binary exponential backoff (BEB) and the fast collision resolution (FCR) scheme. It provides a significant efficient fair sharing performance improvement which converges to the ideal fairness performance while providing high throughput performance.

Low Power Small Area High Performance 2D-DCT architecture

This paper presents an efficient algorithm for implementing the Discrete Cosine Transform (DCT) with Distributed Arithmetic (DA). The proposed algorithm offers high speed and reduced area. The design is free of multipliers, ROM and ROM accumulators. Inner product computational module has been proved mathematically, to require only additions , The number of additions is decreased by dividing the adder array matrix before exploiting the redundancy in it .The proposed architecture uses the recursive DCT algorithm and requires less area than the conventional approaches, 15 additions, 4 subtractions are needed to complete the first phase of the calculations. Hardware implementation of the design is realized on xilinx fpga XC3S1000 .Savings exceed 95%.

A Comparison of Principal Component Analysis and Generalized Hebbian Algorithm for Image Compression and Face Recognition

In this paper we perform image compression and face recognition using principal component analysis (PCA) and the generalized Hebbian algorithm (GHA) which is one of the PCA techniques involving neural network. By implementing the PCA and GHA algorithms for image compression we found that PCA gives better compression ratio to the image than GHA and as for face recognition we found that GHA gives more recognition rate than PCA

Study of LTE — R X2 handover based on A3 event algorithm using MATLAB

Railway wireless communication has become an indispensable part to provide the reliability and safety of the railway operation as well as the improved railway services. GSM for Railway (GSM-R) system is a widely used means of railway communication, its commercial success are fully explained that GSM-R system has maturity and reliability. Along with the developments of the high-speed railway, higher and higher technical requirements for the high-speed railway mobile communication system are raised. GSM-R has shown many limitations in high speed, high-quality transmission. Therefore, long term evolution (LTE), is basically confirmed as the next generation railway wireless communication technology, namely LTE - R. As the fourth generation, LTE - R provides a high transmission speed and high mobility to the user. As trains move with relatively high speed, they suffer from many handovers. In this paper we study and simulate, using MATLAB, handover in LTE - R system. We choose A3 event handover algorithm based on measuring Reference signal Received Power (RSRP). The impact of changing both velocity and transmitted power is also simulated.

New approach for the fault diagnosis of linear analog networks

This paper presents two new theories for the fault diagnosis of the linear analog networks of unknown impedance matrices but of known terminal (port) voltages. The concept of the companion network is proposed to fault diagnose the network under test. The impedance matrices of the nominal network and its companion network constitute two overdetermined systems of equations with the two corresponding network port voltage vectors. The idea is based on the block dependency between the nominal impedance matrix of the network under test and the impedance matrix of its companion network after excluding the circuit element or subnetwork under test. If the companion impedance matrix is consistent with the measured port voltage vector, then, the nominal impedance matrix corresponding to the circuit element or subnetwork under test must also be consistent with the same measured port voltage vector which means that this circuit element is faulty or is one of the entities of the ambiguity fault set. By this way all circuit elements can be tested using the companion network only.

Neural Network Classification for Iris Recognition Using Both Particle Swarm Optimization and Gravitational Search Algorithm

This paper introduces an iris classification system using FFNNGSA and FFNNPSO. This iris identification system consists of localization of the iris region, normalization, feature extraction and then classification as a final stage. A Canny Edge Detection scheme and a Circular Hough Transform are used to detect the iris boundaries. After that the extracted IRIS region is normalized using Daugman rubber sheet model. Next, Haar wavelet transform is used for extracting features from the normalized iris region then the feature matrix is reduced using the principle component analysis (PCA). Finally, both particle swarm optimization (PSO) and gravitational search algorithm (GSA) are used for training a forward neural network to get the optimum weights and biases. The results showed that training the feed-forward neural network by GSA is better than training it by PSO in an iris recognition system.

Fair Delay Optimization-based Resource Allocation Algorithm for Video Traffic over Wireless Multimedia System

The major issue related to the realization of wireless multimedia system is the design of suitable medium access control (MAC) protocol. The design challenge is to maximize the utilization of the limited wireless resources while guaranteeing the various quality of service requirements for all traffic classes especially for the stringent real-time constraint of real time variable bit rate (rt-VBR) video service. In this paper a novel resource allocation algorithm for video traffic is proposed. The proposed allocation algorithm aims to provide fair delay for video packets by minimizing the delay difference among transmitted video packets. At the same time it adaptively controls the allocated resources (bandwidth) for video traffic around the corresponding average bit rate, and has the ability of controlling the quality of service (QoS) offered for video traffic in terms of packet loss probability and average delay. A minimized control overhead of only two bits is needed to increase the utilization efficiency. Simulation results show that the proposed algorithm achieves very high utilization and provides nearly fair delay among video packets. Its efficiency is also investigated under traffic integration condition with voice and data traffic to show that the QoS offered to video traffic does not change in the presence of the highest priority voice traffic while data traffic increases the channel utilization to 98% by using the remaining bandwidth after voice and video traffic while a good QoS is offered to voice and data traffic.

Adaptive watermarking techniques based on multi-scale morphological image segmentation

Two new image adaptive watermarking techniques which are based on multi-scale morphological segmentation are presented in this paper. One of the proposed techniques extends the conventional spatial domain LSB watermarking technique while the other is an extension of blind watermarking in the discrete wavelet domain. The effect of the use of segmentation information on the quality of the watermarked images is discussed and examples are given to show the advantage of morphological segmentation as a preprocessing step for image adaptive watermarking