Majlesi Journal of Electrical Engineering
Volume:16 Issue: 3, Sep 2022

The rapid increase in the number of medical image repositories nowadays has led to problems in managing and retrieving medical visual data. This has proved the necessity of Content-Based Image Retrieval (CBIR) with the aim of facilitating the investigation of such medical imagery. One of the most serious challenges that require special attention is the representational quality of the embeddings generated by the retrieval pipelines. These embeddings should include global and local features to obtain more useful information from the input data. To fill this gap, in this paper, we propose a CBIR framework that utilizes the power of deep neural networks to efficiently classify and fetch the most related medical images with respect to a query image. Our proposed model is based on combining Vision Transformers (ViTs) and Convolutional Neural Networks (CNNs) and learns to capture both the locality and also the globality of high-level feature maps. Our method is trained to encode the images in the database and outputs a ranking list containing the most similar image to the least similar one to the query. To conduct our experiments, an intermodal dataset containing ten classes with five different modalities is used to train and assess the proposed framework. The results show an average classification accuracy of 95.32 % and a mean average precision of 0.61. Our proposed framework can be very effective in retrieving multimodal medical images with the images of different organs in the body.

Analyzing human behavior and detecting visual anomaly are important applications of video surveillance systems in many fields such as security systems and intelligent buildings. Person re-identification (RE-ID) is one of the main steps in a surveillance system, where, it has directly an effect on system performance. Occluded body parts, backgrounds clutter, and variations in pose and scene illumination are some noticeable problems in appearance-based RE-ID approaches, as they have an effect on pedestrians’ appearance during tracking task. Among the mentioned problems, person RE-ID considering appearance changes caused by occlusion can be considered as a common problem in video surveillance systems. In this paper, some existing people RE-ID approaches are briefly reviewed in terms of robustness to person body occlusion. Also, the experimental results reported in these approaches are compared using some partial occluded databases. The comparison results demonstrate the supremacy of the non-pose-guided RE-ID approaches.

High Efficiency Video Coding (HEVC) is considered last standard for video compression with about 50% additional compression compared to the previous standard i.e. H264/AVC, while maintaining image quality. The significant increase in performance of this standard has been achieved with high computational complexity. In this standard, the intra prediction unit is one of the parts that, due to the increase in the number of prediction modes, although it greatly improves performance, significantly increases its computational complexity. In this paper, a method has been proposed to reduce the number of intra prediction modes in HEVC by which the computational complexity of compression at this stage can be reduced as much as possible. The proposed method determines the predominant mode of 4 × 4 blocks, determines the details in each larger block, and accordingly, by applying the appropriate filters and selecting the most likely mode, the number of candidate modes to select the best mode is reduced. The simulation results showed that on average the proposed method can reduce the compression time by 45% while increasing the bit rate by 0.69%.

There can be observed a gradual transition of the electric power industry to the innovative technology platform Smart Grid around the world; in Russia it is done on the platform of an intelligent electrical grid with an active-adaptive network, which becomes one of the most important subsystems. There are new elements of power transmission lines, among which new generation conductors (NGC) are being actively introduced. However, the high cost of innovations in the power grid complex makes their use quite slow. The conductor cross-section is the most significant parameter of the power transmission line; it determines its main technical and economic indicators. The problem of choosing economically reasonable conductor cross-sections stems from the need to ensure the required level of reliability of power transmission lines, determine the most effective way to invest money and reduce the cost of electricity transportation. Modern requirements for the feasibility study of design solutions increase the economic significance of the problem to determine the economically feasible conductor cross-sections for the power transmission lines. At the same time, there is no method for choosing the optimal conductor cross-sections of NGC. An incorrect choice of the conductor brand and its cross-section can lead to unjustified costs for the construction and reconstruction of power transmission lines and increase the cost of electricity transmission. In the article there is the analysis of existing methods of project decisions feasibility; it was taken as the basis to develop the technical and economic model of a conductor cross-section for active-adaptive electrical grids taking into account the thermal model of the conductor and the random nature of the change of current running through the power transmission lines.

In this paper, a 1 * 2 all-optical power splitter has been presented which is suitable for the third window of optical communications based on photonic crystal structures. This structure can provide a 50% transmission coefficient at a wavelength of 1550 nm in each splitter output branch. The device has been designed based on an input waveguide, two output waveguides, and an L4 resonant cavity in which the transmission coefficient of the structure is improved by infiltrating an optical fluid in some holes without any changes in the place and size of the radius of holes.

An Electrocardiogram (ECG) is a test that is done with the objective of monitoring the heart’s rhythm and electrical activity. It is conducted by attaching a specific type of sensor to the subject’s skin to detect the signals generated by the heartbeats. These signals can reveal significant information about the wellness of the subjects’ heart state, and cardiologists use them to detect abnormalities. Due to the prevalence of heart diseases amongst individuals around the globe, there is an urgent need to design computer-aided approaches to automatically analyze ECG signals. Recently, computer vision-based techniques have demonstrated remarkable performance in medical image analysis in a variety of applications and use cases. This paper proposes an approach based on Convolutional Autoencoders (CAEs) and Transfer Learning (TL). Our approach is an ensemble way of learning the most useful features from both the signal itself, which is the input of the CAE, and the spectrogram version of the same signal, which is fed to a convolutional feature extractor named MobileNetV1. Based on the experiments conducted on a dataset collected from 3 well-known hospitals in Baghdad, Iraq, the proposed method claims good performance in classifying four types of problems in the ECG signals. Achieving an accuracy of 97.3% proves that our approach can be remarkably fruitful in situations where access to expert human resources is scarce.

The internet of things (IoT) is a promising expansion of the traditional Internet, which provides the foundation for millions of devices to interact with each other. IoT enables these smart devices, such as home appliances, different types of vehicles, sensor controllers, and security cameras, to share information, and this has been successfully done to enhance the quality of user experience. IoT-based mediums in day-to-day life are, in fact, minuscule computational resources, which are adjusted to be thoroughly domain-specific. As a result, monitoring and detecting various attacks on these devices becomes feasible. As the statistics prove, in the Mirai and Brickerbot botnets, Distributed Denial-of-Service (DDoS) attacks have become increasingly ubiquitous. To ameliorate this, in this paper, we propose a novel approach for detecting IoT malware from the preprocessed binary data using transfer learning. Our method comprises two feature extractors, named ResNet101 and VGG16, which learn to classify input data as malicious and non-malicious. The input data is built from preprocessing and converting the binary format of data into gray-scale images. The feature maps obtained from these two models are fused together to further be classified. Extensive experiments exhibit the efficiency of the proposed approach in a well-known dataset, achieving the accuracy, precision, and recall of 96.31%, 95.31%, and 94.80%, respectively.

In this paper, we compare dispersion characteristics of chloroform-core photonic crystal fibers (PCFs) with circular lattice and hexagonal lattice in the case of different air hole diameters. By varying lattice constant Ʌ and filling factor in the first ring d1/Ʌ, we can easily control chromatic dispersion and achieve three optimal structures for the circular lattice of fibers #CF1 (Ʌ = 1.0 µm, d1/Ʌ = 0.65), #CF2 (Ʌ = 1.0 µm, d1/Ʌ = 0.7), and #CF3 (Ʌ = 2.0 µm, d1/Ʌ = 0.3) and two optimal structures for the hexagonal lattice of fibers #HF1 (Ʌ = 1.0 µm, d1/Ʌ = 0.5), #HF2 (Ʌ = 2.0 µm, d1/Ʌ = 0.3). At the same structural parameter (Ʌ = 2.0 µm, d1/Ʌ = 0.3) and the corresponding pumping wavelength, the circular structure has a dispersion smaller by 5.598 ps/mn/km than the hexagonal lattice. #CF1 has all-normal dispersion with the peak of the dispersion curve asymptote to the zero-dispersion curve which is very suitable for coherent supercontinuum generation (SCG). The #HF1 structure has a near-zero flat anomalous dispersion in the wide wavelength range from 1.1 µm to 1.4 µm. Our results will be an important premise in choosing a PCF structure to study SCG.

Information is the driving force in vehicular ad hoc networks (VANET) since vehicles share information (emergency, general, and multimedia). VANET communicates between vehicles using a unique routing protocol, unlike other wireless routing technologies. Many protocols, techniques, and approaches have been developed to secure and protect data. To enhance current security and privacy measures and develop and model new ones, the ideas of machine learning (ML), deep learning (DL), and artificial intelligence are being applied. In this paper, we provide information on the various types of attacks that target VANET communication, VANET layers, the security goals that are affected, and real-time attacks that occur on manufacturing hubs. We compared various VANET attack prevention, detection, and AI techniques proposed, as well as future research work in the field of VANET, for improving accuracy, security, and privacy

In this study, the Manganese zinc ferrite nanoparticles with diameters less than 50 nm were prepared. By XRD (X–ray diffraction), scanning electron microscopy (SEM) and transmission electron microscopy (TEM) the morphology and the structure of this ferrite were studied. The X-ray analysis shows the formation of manganese zinc ferrite in spinel phase. SEM photograph is shown spherical shape of nanoparticles. And the TEM confirmed the nanoscale dimensions of the samples. The manganese zinc ferrite nanoparticles crystallite sizes, calculated by Debye-Scherer formula, were found near 13 nm. Sensitivity properties of this ferrite are investigated in a totally isolated plexi glass box. By injecting 1 mL of liquid and vapor it we will have 200 ppm concentration of each sample in this box. Then the injected vapored sample in this box is exposed to the ferrite. After this step the conductivity of the ferrite in a closed circuit was changed. By changing the sample type amount of this conductivity were varied. Five gases were tested in this project: ethanol, dimethyl formamid, carbon tetrachloride, acetonitrile and acetone. Among these samples the carbon tetrachloride had the best sensitivity performance.

The most commonly used variable speed wind turbine is based on doubly fed induction generator (DFIG). To control the reactive power of DFIG-based wind turbines, several methods are suggested that controls the reactive power of the DFIG with slow dynamics and considerable ripples. This paper proposes a new control method based on the adaptive reference model which controls the active and reactive powers of DFIG with high dynamics and low ripples. Given that, the proposed technique has proportional-integral (PI), selecting the proper coefficient for PI controller is significant. To overcome this problem, the grey-wolf algorithm is used to optimize the PI coefficients. The results show that the proposed method gives satisfactory performance with lower overshoots and faster dynamic response.

The Cobalt zinc ferrite nanoparticles with diameters less than 20 nm were prepared. By XRD (X-ray diffraction), Scanning Electron Microscopy (SEM) and Transmission Electron Microscopy (TEM) the morphology and the structure of this ferrite were studied. The X-ray analysis shows the formation of manganese zinc ferrite in the spinel phase. SEM photograph is shown the spherical shape of nanoparticles. And the TEM confirmed the nanoscale dimensions of the samples. The cobalt zinc ferrite nanoparticles crystallite sizes, calculated by the Debye-Scherer formula, were found near 13 nm. The sensitivity properties of this ferrite are investigated in a totally isolated plexi glass box. By injecting 1 mL of liquid and vaporizing it, we will have 200 ppm concentration of each sample in this box. Then the injected vapored sample in this box is exposed to the ferrite. After this step, the conductivity of the ferrite in a closed circuit was changed. By changing the sample type, amount of this conductivity was varied. Six gases were tested in this project: ethanol, nitrile alcohol, dimethyl formamide, carbon tetrachloride, acetonitrile, and acetone. Among these samples, the carbon tetrachloride had the best sensitivity performance. Finally, the sensor equation for carbon tetrachloride was extracted by applying different concentrations of it from 20 to 200 ppm.