نشریه مهندسی برق و مهندسی کامپیوتر ایران
سال سیزدهم شماره 1 (پیاپی 36، تابستان 1394)

In this paper, a robust method for face detection from different views using a combination of Gabor filters and neural networks is presented. First, a mathematical equation of Gabor filter is expressed. Then, by examining 75 different filter banks, range of effective parameters values in Gabor filter generation is determined, and finally, the best value for them is specified. The neural network used in this paper is a feed-forward back-propagation multilayer perceptron network. The input vector of the neural network is obtained from the convolution the input image and a Gabor filter with angles π / 2 and the frequency π / 2 in the frequency domain. The proposed method has been tested on 550 image samples from Feret database with simple background and Markus Weber database with complex background, and detection accuracy of them is 98.4% and95%, respectively. Also, the face area has been detected using Viola-Jones algorithm, and then comparison between the results obtained from Viola-Jones algorithm and the proposed method is described.

Today's advanced systems are expected to be able to adapt to environmental conditions and unpredictable situations. The first requirement for such systems is to adjust them according to customer needs, their own ability and operational environment and they should be able to answer when faced with problem and unexpected request. Software adaptation techniques try to cope, with adaptation contracts and reconfiguration capabilities. Also these reconfigurations should be performed out of the sight of client and sometimes during the operation so that prohibit system designers from direct involvement in the internal affairs of clients. Sometimes these adaptation techniques have an impressive role in reusing components for making new systems or improving old ones. Thins paper try to create a system that can be adapted to the environment and besides it also reduces the complexity problem. To do so, at first we use a formal model to represent the whole system and then, build a mathematical model called adaptor based on adaptation contract and client requests. After creation of the adaptor, the all configuration and transactions between the client and system are done through the adaptors and Adaptors are responsible for coordinating the internal system components. Also, to avoid complexity, the concept of hierarchical networks and services are used for building the networks of adaptors.

Spoken Term Detection (STD) approaches can be divided into two main groups: Hidden Markov Model (HMM)-based and Discriminative STD (DSTD) approaches. One of the important advantages of HMM-based methods is that they can use context dependent (diphone or triphones) information to improve the whole STD system performance. On the other hand, lack of triphones information is one of the significant drawbacks of DSTD methods. In this paper, we propose a solution to overcome this drawback of DSTD systems. To this end, we modify the feature extraction part of an Evolutionary DSTD (EDSTD) system to consider triphones information. At first, we propose a monophone-based feature extraction part for the EDSTD system. Then, we propose an approach for exploiting triphones information in the EDSTD system. The results on TIMIT database indicate that the true detection rate of the triphone-based EDSTD (Tph-EDSTD) system, in false alarm per keyword per hour greater than two, is about 3% higher than that of the monophone-based EDSTD (Mph-SDSTD) system. This improvement costs about 36% degradation of the system response speed which is neglected.

The current architecture of data center networks is a combination of Ethernet switches and routers. However, this architecture cannot satisfy the requirements of these networks. Ethernet switches are flexible, have simple configuration, but are not scalable. Routers provide better scalability and efficient use of bandwidth, but are costly. This architecture has a noticeable overhead configuration and maintenance. So, if we had a larger Layer 2 networks, number of routers and consequently the costs will be lessened. Many methods are presented for this purpose.In this paper introduce some main requirement center data networking and characteristic of proposed methods. Among of these methods, OpenFlow is preferred.But the control overhead of OpenFlow is high. One way to reduce the control overhead by separating big and small flows and letting the controller to control only the big flows. ECMP routing is a method that can be used for routing small flows. However OpenFlow does not support ECMP. In this paper, a new method based on OpenFlow is proposed to replace ECMP. The proposed method can achieve performance comparable to ECMP.

In this paper, a method is proposed to automatically select reference image in histogram matching. Histogram matching is one of the simplest spatial image enhancement methods which improves contrast of the initial image based on histogram of the reference image. In the conventional histogram matching methods, user should perform several experiments on various images to find a suitable reference image. This paper presents a new method to automatically select the reference image. In this method, images are converted from RGB to HSV, and the illumination (V) components are considered to select the reference image. The appropriate reference image is selected using a similarity measure via measuring the similarity between the histograms of the initial image and histograms of the images in the data base. Indeed, an image with similar histogram to the histogram of the original images is more appropriate to choose as the reference image for histogram matching. Results in this research indicate superiority of the proposed approach, compared to other existing approaches, in image enhancement via histogram matching. In addition, the user would have no concern in selecting an appropriate reference image for histogram matching in the proposed approach. This approach is applicable to both RGB and gray scale images.

Much attention has recently been paid to methods of shared secret key generation that exploit the random characteristics of the amplitude and phase of a received signal and common channel symmetry in wireless communication systems. Protocols based on the phase of a received signal, due to the uniform distribution phase of fading channel, are suitable in both static and dynamic environments and, they have a key generation rate (KGR) higher than protocols based on received signal strength (RSS).In addition, previous works have generally focused on key generation protocol for single-antenna (SISO) systems but these have not produced a significant KGR. So in this paper to increase the randomness and key generation rate are used received signal phase estimations on multiple-antenna (MIMO) systems because they have the potential to present more random variables in key generation compared to SISO systems. The results of simulation show that the KGR of the proposed protocol is 4 and 9 times more than the KGR of a SISO system, when the numbers of transmitter and receiver antennas are the same and equal to 2 and 3, respectively. Also, the key generation rate will increase considerably, when to extract the secret key bits using multilevel quantization.

One of the problems with traditional genetic algorithms is its premature convergence that makes them incapable of searching good solutions of the problem. A memetic algorithm (MA) which is an extension of the traditional genetic algorithm uses a local search method to either accelerate the discovery of good solutions, for which evolution alone would take too long to discover, or to reach solutions that would otherwise be unreachable by evolution or a local search method alone. In this paper, a memetic algorithm based on learning automata (LA) and memetic algorithm, called LA-MA, is introduced. This algorithm is composed of two parts, genetic section and memetic section. Evolution is performed in genetic section and local search is performed in memetic section. The basic idea of LA-MA is to use learning automata during the process of searching for solutions in order to create a balance between exploration performed by evolution and exploitation performed by local search. To evaluate the efficiency of LA-MA, it has been used to solve two optimization problems: OneMax and graph isomorphism problems. The results of computer experimentations have shown that different versions of LA-MA outperform the others in terms of quality of solution and rate of convergence.

The most relevant method to detect epileptic seizures is the electroencephalogram (EEG) based signal processing method which, due to the need for installing some electrodes on different places of the person's head, causes many movement problems. The aim of this research is to automatically and intelligently detect grand-mal epileptic seizures and also to recognize normal activities of a person suffering from the disease by video surveillance. In this paper we have used the combination of machine vision and machine learning techniques to automatically detect grand-mal epileptic seizure when the person is lying on the ground or on the bed. After subtracting the background from video frame sequences and extracting the image silhouette, appropriate geometrical features have been extracted and fed to the multi-class support vector machine as the input for automatically classifying the videos and assigning proper activity label. All the implementations have been done on MATLAB R2011a. In this intelligent system the accuracy of detecting and recognizing activities is 90.21%. Using this system in addition to reducing the number of human observers is very helpful for the on time and constant detection of the condition. The need for just a conventional video camera and a computer system makes it affordable for people with different incomes. Because it needs not to be in contact with the person's body, there is no movement problem too. High accuracy verifies the optimal performance of the system.

In many applications, wireless mesh networks work by battery as a power source. In this scenario, routing method has a great impact on the network lifetime.In this research a new backbone based wireless mesh network routing method for maximizing lifetime has been proposed. This approach is compatible with the features provided by IEEE standard for wireless mesh networks. In this method, backbone routers are selected based on the maximum remaining energy.The proposed algorithm is compared with optimum and shortest path routing methods. Simulation results show acceptable increase in network lifetime in the proposed approach.