Research Interests:
In this work, we present an effective method for automatic Arabic Sign Language recognition that uses a Convolutional Neural Network (CNN) for feature extraction and a Long Short-Term Memory (LSTM) for classification. AlexNet, a CNN... more
In this work, we present an effective method for automatic Arabic Sign Language recognition that uses a Convolutional Neural Network (CNN) for feature extraction and a Long Short-Term Memory (LSTM) for classification. AlexNet, a CNN architecture, is used to extract deep features from the input image while the LSTM is used to preserve the sequential structure of the video frames. The method was tested on a data set consisting of 50 repetitions of 150 signs commonly used in daily activities performed by three signers. The proposed method achieved an overall recognition accuracy of 95.9% for the signer-dependent case, and 43.62% for the more difficult signer-independent case.
Research Interests:
Research Interests:
Accurate wind speed prediction is important for wind energy integration into the power grid. While most wind turbines have hub heights of about 80 - 140m, wind speeds are usually measured up to 40mm and in exceptional cases up to 100m.... more
Accurate wind speed prediction is important for wind energy integration into the power grid. While most wind turbines have hub heights of about 80 - 140m, wind speeds are usually measured up to 40mm and in exceptional cases up to 100m. This paper analyzes the predictability of wind speed with heights. To achieve this, a Laser Illuminated Detection and Ranging (LiDAR) system, ZephIR 300, was acquired and installed at the beach of King Fahd University of Petroleum & Minerals. The ZephIR 300 device is widely accepted for wind resource assessment and its wind speed measurements have been validated and found to be accurate for heights from 10 to 300m. Wind speed data was collected at 20, 40, 50, 60, 80, 100, 120, 140, 160, and 180m heights for three months. The collected data was used for training and testing the performance of the RNN model for predicting the wind speed 12 hours ahead of time using 48 previous hourly values. Careful analyses of short-term wind speed prediction at different heights and future hours showed that wind speed is predicted more accurately at higher heights. For example, the mean absolute percent error decreased from 0.15 to 0.11 corresponding to heights 20 and 180m; respectively.
Research Interests:
Sign language is important for facilitating communication between hearing impaired and the rest of society. However, most vocal people do not understand sign language, hence, the need to develop system capable of translating sign... more
Sign language is important for facilitating communication between hearing impaired and the rest of society. However, most vocal people do not understand sign language, hence, the need to develop system capable of translating sign language. Two approaches have traditionally been used in the literature: image-based and glove-based systems. Glove-based systems require the user to wear electronic gloves while performing the signs. The glove includes a number of sensors detecting different hand and finger articulations. Image-based systems use camera(s) to acquire a sequence of images of the hand. Each of the two approaches has its own disadvantages. The glove-based method is not natural as the user must wear a cumbersome instrument while the camera-based system requires specific background and environmental conditions to achieve high accuracy. In this paper, we propose a new approach for Arabic Sign Language Recognition (ArSLR) which involves the use of two Leap Motion Controllers (LMC) to prevent the case of one finger being occluded by another finger or hand. This device detects and tracks the hand and fingers to provide position and motion information. We propose to use the two LMCs as a backbone of the ArSLR system. In addition to data acquisition, the system includes a preprocessing stage, a feature extraction stage, and a classification stage. Fusion of evidences from the two LMCs at the feature extraction and classification stage was also investigated using Dempster-Shafer theory of evidence. Features fusion from the two LMCs gives 97.7% classification accuracy with Linear Discriminant Analysis (LDA) classifier and 97.1% with classifier level fusion. This gives better recognition over the use of a single LMC.
Research Interests:
Images and videos are seen as the most reliable source of visual information as they are a fundamental part of the multimedia world. For instance, face recognition technology utilizes images for security purposes. However, due to either... more
Images and videos are seen as the most reliable source of visual information as they are a fundamental part of the multimedia world. For instance, face recognition technology utilizes images for security purposes. However, due to either the physical properties of the acquisition equipment (internal) or the nature of the environment (external), images can be affected by a wide range of distortions. Researchers have enumerated more than 24 distortions that can affect images. Among which four types are the most prominent ones. In this paper, a novel no-reference color image quality assessment technique is introduced. The technique is based on extracting a set of features from the High Order Singular Value Decomposition (HOSVD) of images. Such features are then used with a neural network regressor to predict the quality score. The results show excellent performance exceeding traditional techniques based only on gray-scale images.
Research Interests:
In the Kingdom of Saudi Arabia, sandstorms are quite frequent and cause dust accumulation on PV panel surfaces that act as a barrier to solar radiation. This decreases of the solar radiation energy absorption and subsequently reduces the... more
In the Kingdom of Saudi Arabia, sandstorms are quite frequent and cause dust accumulation on PV panel surfaces that act as a barrier to solar radiation. This decreases of the solar radiation energy absorption and subsequently reduces the energy output of the panels. The present effort aims at reducing the dust accumulation on PV panels by flying the drone above these panels at certain heights and time intervals. This paper demonstrates the effectiveness of a drone flying over photovoltaic (PV) panels to remove accumulated dust and improve their efficiency. The downward thrust of the drone due to its cruise at a certain height above the PV panels is able to remove most of the accumulated dust if performed regularly. The tests were conducted at King Fahd University of Petroleum and Minerals (KFUPM) beach, Dhahran, Saudi Arabia by loading each panel uniformly with 20, 50, and 100 CC of dust.
Sign language is the major means of communication for the deaf community. It uses body language and gestures such as hand shapes, lib patterns, and facial expressions to convey a message. Sign language is geography-specific, as it differs... more
Sign language is the major means of communication for the deaf community. It uses body language and gestures such as hand shapes, lib patterns, and facial expressions to convey a message. Sign language is geography-specific, as it differs from one country to another. Arabic Sign language is used in all Arab countries. The availability of a comprehensive benchmarking database for ArSL is one of the challenges of the automatic recognition of Arabic Sign language. This article introduces KArSL database for ArSL, consisting of 502 signs that cover 11 chapters of ArSL dictionary. Signs in KArSL database are performed by three professional signers, and each sign is repeated 50 times by each signer. The database is recorded using state-of-art multi-modal Microsoft Kinect V2. We also propose three approaches for sign language recognition using this database. The proposed systems are Hidden Markov Models, deep learning images’ classification model applied on an image composed of shots of the...
Research Interests:
The objective of this work is to understand the fluctuating nature of wind speed characteristics on different time scales and to find the long-term annual trends of wind speed at different locations in South Africa. The hourly average... more
The objective of this work is to understand the fluctuating nature of wind speed characteristics on different time scales and to find the long-term annual trends of wind speed at different locations in South Africa. The hourly average mean wind speed values over a period of 20 years are used to achieve the set objective. Wind speed frequency, directional availability of maximum mean wind speed, total energy, annual energy yield and plant capacity factors are determined for seven locations situated both inland and along the coast of South Africa. The highest mean wind speed (6.01 m/s) is obtained in Port Elizabeth and the lowest mean wind speed (3.86 m/s) is obtained in Bloemfontein. Wind speed increased with increasing latitudes at coastal sites (Cape Town, Durban, East London and Port Elizabeth), while the reverse trend was observed at inland locations (Bloemfontein, Johannesburg and Pretoria). Noticeable annual changes and relative wind speed values are found at coastal locations compared to inland sites. The energy pattern factor, also known as the cube factor, varied between a minimum of 1.489 in Pretoria and a maximum of 1.858 in Cape Town. Higher energy pattern factor (EPF) values correspond to sites with fair to good wind power potential. Finally, Cape Town, East London and Port Elizabeth are found to be good sites for wind power deployments based on the wind speed and power characteristics presented in this study.
Research Interests:
... 104-1 1 1. [2] EMC, Wong, phone-based remote controller for home and office automation, IEEE Transactions on Consumer ... [4] Liang, Li-Chen Fu and Chao-Lin W, An integrated, flexible, and Internet-based control architecture for... more
... 104-1 1 1. [2] EMC, Wong, phone-based remote controller for home and office automation, IEEE Transactions on Consumer ... [4] Liang, Li-Chen Fu and Chao-Lin W, An integrated, flexible, and Internet-based control architecture for home automation system in the ...
Research Interests:
ABSTRACT
Research Interests:
ABSTRACT We consider the problem of over-the-horizon multiradar track association. The operation center receives the registration of tracks from three different over-the-horizon radars. The fusion of these tracks presents a large operator... more
ABSTRACT We consider the problem of over-the-horizon multiradar track association. The operation center receives the registration of tracks from three different over-the-horizon radars. The fusion of these tracks presents a large operator workload. We investigate automating this process. We propose two methods for feature extraction. The first uses the Hough transform and the second uses some track affinity measures. For every case, we consider two systems, one using the multilayer perceptron and the other assuming independence of the features and combining the probabilities. For the Hough transform, we consider the range-time image of the tracks and transform all track data to a unified coordinate system. In the second method, we consider a set of affinity measures and study their relative and cumulative frequencies for associated and nonassociated pairs of tracks. Error rates of less than 7% were achieved with the Hough transform method and less than 1% using track affinity measures.
Research Interests:
Research Interests: Computer Science, Distributed Computing, Sign Language, Face Detection, Region growing, and 9 moreComputer Software, hidden Markov model, Sign language recognition, Gaussian Mixture, Skin Color, Reference Point, Electrical And Electronic Engineering, Time varying, and Computers and Electrical Engineering
ABSTRACT This paper describes a system for tracking and identifying pilgrims in the Holy areas in Makkah, Saudi Arabia and the surroundings, during Hajj season (Pilgrimage) using the mobile phone. It utilizes an internet enabled 3.5 G... more
ABSTRACT This paper describes a system for tracking and identifying pilgrims in the Holy areas in Makkah, Saudi Arabia and the surroundings, during Hajj season (Pilgrimage) using the mobile phone. It utilizes an internet enabled 3.5 G mobile wireless network, as service providers have already covered the Holy area with such networks. The performance of the developed system is compared with a system that uses a Wireless Sensor Network and is interfaced to the internet through a gateway available from a service provider. The developed system uses a regular mobile phone equipped with a Global Positioning System (GPS) that is available with most pilgrims, while in the second system a small and light weight mobile sensor unit is given to each pilgrim to be fixed on his Hajj cloths. The sensor unit includes a GPS chip, a microcontroller, antennas, and a battery. A network of fixed master units is installed throughout the Holy area for receiving and forwarding data. In both systems, upon request or periodically, the mobile sensor unit or the mobile phone sends its User IDentification number, latitude, longitude, and a time stamp. A server maps the location information on a Google map or a similar geographical information system. If the internet connection is lost, the mobile unit stores the location information in its memory. Once the internet connection is restored, the mobile unit sends all stored location information and clears this information from the memory. Both systems can be used to track a specific pilgrim. Alternatively, any pilgrim can request emergency help. Identification of pilgrims can be achieved by a separate Near Field Communication (NFC) tags if the mobile phone is not NFC enabled. Pilot experiments were carried out successfully during the recent pilgrimage seasons.
Research Interests:
This paper develops a new framework for data compression in seismic sensor networks by using the distributed principal component analysis (DPCA). The proposed DPCA scheme compresses all seismic traces in the network at the sensor level.... more
This paper develops a new framework for data compression in seismic sensor networks by using the distributed principal component analysis (DPCA). The proposed DPCA scheme compresses all seismic traces in the network at the sensor level. First of all, the statistics of the seismic traces acquired at all sensors are represented by a mixture model of a number of probability density functions. Based on this mixture model, the DPCA finds the global PCs at the fusion center. These PCs are then sent back to all sensors so that each sensor projects its own traces over these PCs. This scheme does not require transmitting the original traces, here, leading to a low computational load and a high compression ratio, compared with compression obtained using the local PC analysis (LPCA). Furthermore, we develop an efficient communication solution for the DPCA implementation on practical sensor networks. Finally, the proposed scheme is evaluated using real and synthetic seismic data showing improved performance over the LPCA and the traditional 2-D discrete cosine transform (DCT-2-D) compression. Specifically, to preserve a given signal energy during the compression, the DPCA is shown to achieve a higher compression ratio than the LPCA and the DCT-2-D.
Research Interests:
a series of events addressing the fundamentals advanced scientific computing and specific mechanisms and algorithms for particular sciences. The conference provided a forum where researchers were able to present recent research results... more
a series of events addressing the fundamentals advanced scientific computing and specific mechanisms and algorithms for particular sciences. The conference provided a forum where researchers were able to present recent research results and new research problems and directions related to them. With the advent of high performance computing environments, virtualization, distributed and parallel computing, as well as the increasing memory, storage and computational power, processing particularly complex scientific applications and voluminous data is more affordable. With the current computing software, hardware and distributed platforms effective use of advanced computing techniques is more achievable. The event was very competitive in its selection process and very well perceived by the international scientific and industrial communities. As such, it has attracted excellent contributions and active participation from all over the world. We were very pleased to receive a large amount of...
a multi-track event covering a large spectrum of topics related to advanced engineering computing and applications in sciences. With the advent of high performance computing environments, virtualization, distributed and parallel... more
a multi-track event covering a large spectrum of topics related to advanced engineering computing and applications in sciences. With the advent of high performance computing environments, virtualization, distributed and parallel computing, as well as the increasing memory, storage and computational power, processing particularly complex scientific applications and voluminous data is more affordable. With the current computing software, hardware and distributed platforms effective use of advanced computing techniques is more achievable. The goal of ADVCOMP 2015 was a forum to bring together researchers from the academia and practitioners from the industry in order to address fundamentals of advanced scientific computing and specific mechanisms and algorithms for particular sciences. The conference provided a forum where researchers were able to present recent research results and new research problems and directions related to them. The conference sought contributions presenting nove...
Seismic surveys consist of large volumes of data acquired by a network of sensors. Such survey data is usually corrupted by different types of noise and other distortions. Under these conditions, the recovery of correct seismic amplitudes... more
Seismic surveys consist of large volumes of data acquired by a network of sensors. Such survey data is usually corrupted by different types of noise and other distortions. Under these conditions, the recovery of correct seismic amplitudes has become more challenging. The accuracy of such amplitudes is of primary importance in the pipeline of seismic interpretation. In this work, we introduce a seismic image denoising approach based on the Bilateral filter with Elliptic Gaussian kernel (BFEGK). The Bilateral filter is a non-linear filter that preserves edges and reduces noise. We enhance the filter for seismic image denoising by formulating an elliptic Gaussian kernel. We utilize the interquartile range to obtain a robust estimate of the noise standard deviation. Under Gaussian noise scenarios, the performance is compared to dictionary learning (DL) based denoising, standard Bilateral Filter with Noise Thresholding (BFMT), and Wavelet thresholding (WT) methods using real seismic images. Our proposed method exhibits the closest performance to the DL based denoising compared to GBF and WT methods with a significantly reduced computational load.
Research Interests:
Research Interests:
ABSTRACT Sign language is important for facilitating commu-nication between hearing impaired and the rest of society. Two approaches have traditionally been used in the literature: image-based and sensor-based systems. Sensor-based... more
ABSTRACT Sign language is important for facilitating commu-nication between hearing impaired and the rest of society. Two approaches have traditionally been used in the literature: image-based and sensor-based systems. Sensor-based systems require the user to wear electronic gloves while performing the signs. The glove includes a number of sensors detecting different hand and finger articulations. Image-based systems use camera(s) to acquire a sequence of images of the hand. Each of the two approaches has its own disadvantages. The sensor-based method is not natural as the user must wear a cumbersome instrument while the image-based system requires specific background and environmental conditions to achieve high accuracy. In this paper, we propose a new approach for Arabic Sign Language Recognition (ArSLR) which involves the use of the recently introduced Leap Motion Controller (LMC). This device detects and tracks the hand and fingers to provide position and motion information. We propose to use the LMC as a backbone of the ArSLR system. In addition to data acquisition, the system includes a preprocessing stage, a feature extraction stage, and a classification stage. We compare the performance of Multilayer Perceptron (MLP) neural networks with the Nave Bayes classifier. Using the proposed system on the Arabic sign alphabets gives 98% classification accuracy with the Nave Bayes classifier and more than 99% using the MLP. Keywords—Arabic sign langauge recognition; leap motion con-troller; finger articulation, electronic glove, image-based system.
Research Interests:
Sign language is the basic means of communication among hearing-impaired people. Systems that could act as interpreters between vocal and hearing-impaired people would facilitate the life of deaf and integrate them in the society. Such... more
Sign language is the basic means of communication among hearing-impaired people. Systems that could act as interpreters between vocal and hearing-impaired people would facilitate the life of deaf and integrate them in the society. Such systems should perform bidirectional translation of sign language and spoken language. A lot of efforts invested on translating sign languages to spoken languages. However, it is just as important to translate a spoken language to a sign language. This would provide a two-way communication between hearing-impaired and vocal people. A reasonably priced portable computer would be sufficient to perform the two-way translation. This paper proposes a system that translates Arabic text to Arabic sign language. A word that corresponds to a sign from the Arabic sign language dictionary calls a pre-recorded video clip showing the sign played on the monitor of the portable computer. If the word does not have a corresponding sign in the dictionary, it is finger ...
Research Interests:
Research Interests:
ABSTRACT Support vector machine is proposed to find wind speed at higher heights using measurements at lower heights. The mean absolute percentage error between measured and the estimated wind speed at height 40 m is found to be... more
ABSTRACT Support vector machine is proposed to find wind speed at higher heights using measurements at lower heights. The mean absolute percentage error between measured and the estimated wind speed at height 40 m is found to be satisfactory. After validation at 40 m, the model was used to calculate the wind speed at hub heights up to 100 m. Annual energy yield was found to be increasing with hub height and, hence, accurate estimation of wind speed at heights becomes essential for realistic wind energy assessment. Furthermore, the plant capacity factor was found to be increasing approximately 1% for each 10-m increase in hub height.
Research Interests:
This paper introduces a hybrid system for vehicle access control using RFID and automatic license plate recognition (ALPR) technologies. RFID technology is proven to provide an effective solution to different tracking and localization... more
This paper introduces a hybrid system for vehicle access control using RFID and automatic license plate recognition (ALPR) technologies. RFID technology is proven to provide an effective solution to different tracking and localization problems. However, the technology has its shortcomings in tracking objects/users without a tag. As such, we propose to complement this technology with ALPR to control the access of different types of vehicles to the area of Makkah (Saudi Arabia) during Pilgrimage seasons. This limited area can easily get congested with the huge number of vehicles attempting access. Before the start of the season, vehicles authorized to access the region are assigned passive RFID tags specifying their allowed schedule of entry. Violating vehicles that do not have RFID tags are detected and identified using ALPR. The developed system was tested over two pilgrimage seasons. The experiments showed that the developed RFID system was able to identify all passing vehicles with speeds up to 100 km/h, while the ALPR system achieved 94 % recognition accuracy of vehicles not equipped with RFID tags.
Research Interests:
ABSTRACT
Research Interests:
ABSTRACT The dehumidification process involves simultaneous heat and mass transfer and reliable transfer coefficients are required in order to analyze the system. This has been proved to be difficult and many assumptions are made to... more
ABSTRACT The dehumidification process involves simultaneous heat and mass transfer and reliable transfer coefficients are required in order to analyze the system. This has been proved to be difficult and many assumptions are made to simplify the analysis. The present research proposes the use of ANN based model in order to simulate the relationship between inlet and outlet parameters of the dehumidifier. For the analysis, randomly packed dehumidifier with lithium chloride as the liquid desiccant is chosen. A multilayer ANN is used to investigate the performance of dehumidifier. For training ANN models, data is obtained from analytical equations. Eight parameters are used as inputs to the ANN, namely: air and desiccant flow rates, air and desiccant inlet temperatures, air inlet humidity, desiccant inlet concentration, dimensionless temperature ratio, and inlet temperature of the cooling water. The outputs of the ANN are the water condensation rate and the outlet desiccant concentration as well as its temperature. ANN predictions for these parameters are validated well with experimental values available in the literature with R2 value in the range of 0.9251–0.9660. This study shows that liquid desiccant dehumidification system can be alternatively modeled using ANN with a reasonable degree of accuracy.Research highlights► Artificial neural network (ANN) based model is used to simulate the performance of the liquid desiccant dehumidification process. ► Three ANNs each with eight inputs and one output have been trained. ► Water condensation rate, outlet desiccant concentration and its temperature are predicted. ► ANNs predicted parameters are validated well with the experimental results.
Research Interests:
A substructure-based neural network is proposed for the active control of flexible structures. A flexible structure is divided into substructures. Subsequently, subcontrollers are designed for these substructures using the linear... more
A substructure-based neural network is proposed for the active control of flexible structures. A flexible structure is divided into substructures. Subsequently, subcontrollers are designed for these substructures using the linear quadratic regulator (LQR) control method. These subcontrollers are assembled to obtain the central feedback controller for the whole structure. A radial basis function neural network is trained to emulate the behavior of this central controller designed from substructure levels. The training is based only on the outputs of sensors collocated with the actuators. Therefore, two distinct advantages of the proposed neural network controller are noted as its training being based on substructural LQR controller and collocated sensor data. The performance of the neural network controller is compared favorably with the complete structural LQR controller for various input forces acting on a large flexible structure.
Research Interests:
ABSTRACT This paper presents a new approach based on artificial neural networks (ANNs) to determine the vapor pressure of three widely used inorganic desiccant solutions, namely, calcium chloride, lithium chloride, and lithium bromide.... more
ABSTRACT This paper presents a new approach based on artificial neural networks (ANNs) to determine the vapor pressure of three widely used inorganic desiccant solutions, namely, calcium chloride, lithium chloride, and lithium bromide. The vapor pressure of liquid desiccants depends on temperature and concentration. Empirical expressions generally provide vapor pressure with limited accuracy. Further, the expressions currently in use are tedious and valid for narrow ranges and must be adjusted constantly. In this paper neural networks were trained to predict vapor pressure of desiccant solutions with a reasonable accuracy without mathematical formulae. Trained neural network models provided wide ranges of vapor pressure for desiccant solutions without the need to cross reference several tables or charts. Results showed potential of using ANNs for the prediction of vapor pressure of desiccant solution for cooling applications.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Abstract In this paper, artificial neural networks (ANN), which are known for their ability to model nonlinear systems and their inherent noise-filtering abilities, are used as O2 analyzer to predict O2 contents in a boiler at SHARQ... more
Abstract In this paper, artificial neural networks (ANN), which are known for their ability to model nonlinear systems and their inherent noise-filtering abilities, are used as O2 analyzer to predict O2 contents in a boiler at SHARQ petrochemical,company,in Saudi Arabia. The training data has been collected over duration of one month,and used to train a neural network to develop neural
Research Interests:
We propose a new framework for evaluating image enhancement methods. In this work, we focus on contrast enhancement, and use the joint probability and the mutual information derived from the image co-occurrence matrix to determine the... more
We propose a new framework for evaluating image enhancement methods. In this work, we focus on contrast enhancement, and use the joint probability and the mutual information derived from the image co-occurrence matrix to determine the proposed index. We show that the second order entropy and the mutual information can be used jointly for tracking and evaluating the effect of contrast enhancement. More importantly, we show that the variations of the mutual information mimic, in a consistent way, human perceptual sensitivity to visual appearance of image changes due to contrast enhancement. The proposed approach is seen as a completely new framework for evaluating image enhancement methods, validated by a number of experiments. Finally, we also show that the proposed metric can even be used to detect image enhancement thresholds at which unpleasant artifacts start to be perceived by viewers.