Research Interests:
Copying and pasting a patch of an image to hide or exaggerate something in a digital image is known as a copy-move forgery. Copy-move forgery detection (CMFD) is hard to detect because the copied part image from a scene has similar... more
Copying and pasting a patch of an image to hide or exaggerate something in a digital image is known as a copy-move forgery. Copy-move forgery detection (CMFD) is hard to detect because the copied part image from a scene has similar properties with the other parts of the image in terms of texture, light illumination, and objective. The CMFD is still a challenging issue in some attacks such as rotation, scaling, blurring, and noise. In this paper, an approach using the convolutional neural network (CNN) and k-mean clustering is for CMFD. To identify cloned parts candidates, a patch of an image is extracted using corner detection. Next, similar patches are detected using a pre-trained network inspired by the Siamese network. If two similar patches are not evidence of the CMFD, the post-process is performed using k-means clustering. Experimental analyses are done on MICC-F2000, MICC-F600, and MICC-F8 databases. The results showed that using the proposed algorithm we can receive a 94.13%...
Research Interests:
Video-based Intelligent Transportation Systems (V-ITS) can play an important role in developing a wide range of applications in transportation field. These systems use the outputs of video cameras to extract desired information by the... more
Video-based Intelligent Transportation Systems (V-ITS) can play an important role in developing a wide range of applications in transportation field. These systems use the outputs of video cameras to extract desired information by the means of various Artificial Intelligence techniques. Considering impressive advantages of applying Deep Neural Networks (DNNs) in different fields of object detection and classification, these methods have attracted a huge attention among researchers in recent years. In this regard, Convolutional Neural Networks (CNNs) as an important class of DNNs have been used for visual imagery goals in a wide variety of applications such as image recognition and video analysis, and even made their way through ITS applications. One of the most important steps of V-ITS applications is the process of vehicle detection in video frames and the high accuracy rate in this step can provide applicable data for other complementary modules such as vehicle tracking and classification. In this paper, a robust method to detect vehicles in video frames based on CNNs is proposed which provides an almost real-time performance and impressive accuracy. To overcome the challenges of building a precise vehicle detection model from still images, we have transformed the main architecture of a pre-trained ResNet-50 residual network to Faster Region-based Convolutional Neural Network (Faster R-CNN). Experimental results show that the system's sensitivity factor is 0.985 and it needs an average of 74 milliseconds to detect vehicles in real condition data. Consequently, our method can provide acceptable results in vehicle detection in terms of accuracy and execution time.
Research Interests:
In vision-driven Intelligent Transportation Systems (ITS) where cameras play a vital role, accurate detection and re-identification of vehicles are fundamental demands. Hence, recent approaches have employed a wide range of algorithms to... more
In vision-driven Intelligent Transportation Systems (ITS) where cameras play a vital role, accurate detection and re-identification of vehicles are fundamental demands. Hence, recent approaches have employed a wide range of algorithms to provide the best possible accuracy. These methods commonly generate a vehicle detection model based on its visual appearance features such as license plate, headlights, or some other distinguishable specifications. Among different object detection approaches, Deep Neural Networks (DNNs) have the advantage of magnificent detection accuracy in case a huge amount of training data is provided. In this paper, a robust approach for license plate detection (LPD) based on YOLO v.3 is proposed which takes advantage of high detection accuracy and real-time performance. The mentioned approach can detect the license plate location of vehicles as a general representation of vehicle presence in images. To train the model, a dataset of vehicle images with Iranian ...
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
In order to provide the best performance for memory accesses in the multimedia extensions that load or store consecutive subwords from/to memory, the mem- ory access must be correctly aligned. That means that an n-byte transfer must be... more
In order to provide the best performance for memory accesses in the multimedia extensions that load or store consecutive subwords from/to memory, the mem- ory access must be correctly aligned. That means that an n-byte transfer must be set on an n-byte boundary. In most SIMD architectures, unaligned memory accesses have a large performance penalty or are even disallowed. For example, our result shows that for addition of two arrays of size 1024 1024, whose addresses are either aligned or unaligned, aligned code is 1.47 times faster than unaligned code using SSE instructions. Hence, in this paper we evalu- ate the advantages and disadvantages of different techniques to avoid misaligned memory accesses such as replication of data in memory, padding of data structures, loop peeling, and shift instructions. Our result shows that the MMX im- plementation of the FIR filter using replication of data is up to 2.20 times faster than the MMX implementation with mis- aligned accesses. Furtherm...
Research Interests:
The efficient processing of MultiMedia Appli- cations (MMAs) is currently one of the main bottlenecks in the media processing field. Many architectures have been proposed for processing MMAs such as VLIW, superscalar (general-purpose... more
The efficient processing of MultiMedia Appli- cations (MMAs) is currently one of the main bottlenecks in the media processing field. Many architectures have been proposed for processing MMAs such as VLIW, superscalar (general-purpose processor enhanced with a multimedia ex- tension such as MMX), vector architectures, SIMD architec- tures, and reconfigurable computing devices. The question then arises: which architecture can exploit the character- istic features of MMAs the most? In this paper, first, we explain the characteristics of MMAs, after that we discuss the different architectures that have been proposed for pro- cessing MMAs. Subsequently, they are compared based on
Research Interests:
In vision-driven Intelligent Transportation Systems (ITS) where cameras play a vital role, accurate detection and re-identification of vehicles are fundamental demands. Hence, recent approaches have employed a wide range of algorithms to... more
In vision-driven Intelligent Transportation Systems (ITS) where cameras play a vital role, accurate detection and re-identification of vehicles are fundamental demands. Hence, recent approaches have employed a wide range of algorithms to provide the best possible accuracy. These methods commonly generate a vehicle detection model based on its visual appearance features such as license-plate, headlights or some other distinguishable specifications. Among different object detection approaches, Deep Neural Networks (DNNs) have the advantage of magnificent detection accuracy in case a huge amount of training data is provided. In this paper, a robust approach for license-plate detection based on YOLO v.3 is proposed which takes advantage of high detection accuracy and real-time performance. The mentioned approach can detect the license-plate location of vehicles as a general representation of vehicle presence in images. To train the model, a dataset of vehicle images with Iranian license...
Research Interests:
Research Interests:
Skin color based tracking techniques often assume a static skin color model obtained either from an offline set of library images or the first few frames of a video stream. These models can show a weak performance in presence of changing... more
Skin color based tracking techniques often assume a static skin color model obtained either from an offline set of library images or the first few frames of a video stream. These models can show a weak performance in presence of changing lighting or imaging conditions. We propose an adaptive skin color model based on the Gaussian mixture model to handle the changing conditions. Initial estimation of the number and weights of skin color clusters are obtained using a modified form of the general Expectation maximization algorithm, The model adapts to changes in imaging conditions and refines the model parameters dynamically using spatial and temporal constraints. Experimental results show that the method can be used in effectively tracking of hand and face regions.
Research Interests: Computer Science, Artificial Intelligence, Computer Vision, Adaptation, Environmental Change, and 14 moreFace Detection, Video Streaming, Tracking, Motion Detection, Segmentation, Temporal Constraints, Gaussian Mixture Model, Color Space, Skin Color Detection, Mixture Model, Roy Adaptation Model, Skin Color, Expectation Maximization (EM) Algorithm, and Pixel
Research Interests:
Research Interests: Computer Science, Real Time Computing, Intelligent Transportation Systems, Intelligent Transport System, GPGPU (General Purpose GPU) Programming, and 5 moreOpenCL on GPUs, Vehicle Detection, Intelligent Transportation Systems (ITS), Parallel Programming on GPU using CUDA and OpenCL, and Vehicle Speed Detection
Research Interests:
Vehicle counting process provides appropriate information about traffic flow, vehicle crash occurrences and traffic peak times in roadways. An acceptable technique to achieve these goals is using digital image processing methods on... more
Vehicle counting process provides appropriate information about traffic flow, vehicle crash occurrences and traffic peak times in roadways. An acceptable technique to achieve these goals is using digital image processing methods on roadway camera video outputs. This paper presents a vehicle counter-classifier based on a combination of different video-image processing methods including object detection, edge detection, frame differentiation and the Kalman filter. An implementation of proposed technique has been performed using C++ programming language. The method performance for accuracy in vehicle counts and classify was evaluated, which resulted in about 95 percent accuracy for classification and about 4 percent error in vehicle detection targets.
Research Interests:
Research Interests: Computer Science, Distributed Computing, Parallel Computing, New Media, Parallel Programming, and 9 moreComputer Software, Key words, Performance Improvement, Discrete Cosine Transform, SIMD, Multimedia Application, Register File, Single Instruction Multiple Data (SIMD), and Instruction Set Extension
Research Interests:
Research Interests:
Research Interests:
Research Interests:
There are generally two algorithms to traverse an image to implement the 2D Discrete Wavelet Transform (DWT), namely Row-Column Wavelet Transform (RCWT) and Line-Based Wavelet Transform (LBWT). In the RCWT algorithm, the 2D DWT is divided... more
There are generally two algorithms to traverse an image to implement the 2D Discrete Wavelet Transform (DWT), namely Row-Column Wavelet Transform (RCWT) and Line-Based Wavelet Transform (LBWT). In the RCWT algorithm, the 2D DWT is divided into two 1D DWT: hor- izontal and vertical ltering. The horizontal ltering pro- cesses the rows of the original image and stores the wavelet coecien ts in an auxiliary matrix. Thereafter, the verti- cal ltering phase processes the columns of the auxiliary matrix and stores the results back in the original matrix. In the LBWT algorithm, the vertical ltering is started as soon as a sucien t number of rows, as determined by the lter length, has been horizontally processed. In this pa- per, we provide answers to the following questions: rst, which implementation is easier to vectorize using SIMD in- structions? Second, which SIMD implementation provides more performance? Our initial results for Daubechies' transform with four coecien ts show that ...
Research Interests:
Research Interests:
Research Interests: Information Systems, Computer Science, Distributed Computing, Grid Computing, High Performance Computing, and 7 moreGray Level Co-occurrence Matrix, Future Generation Comp. Syst., High performance computer, Processing Element, General Purpose Processor (GPP), Reconfigurable Processor, and Independent set
Research Interests:
Today’s hardware platforms have parallel processing capabilities and many parallel programming models have been developed. It is necessary to research an efficient implementation of compute-intensive applications using available... more
Today’s hardware platforms have parallel processing capabilities and many parallel programming models have been developed. It is necessary to research an efficient implementation of compute-intensive applications using available platforms. Dense matrix-matrix multiplication is an important kernel that is used in many applications, while it is computationally intensive, especially for large matrix sizes. To improve the performance of this kernel, we implement it on the graphics processing unit (GPU) platform using the tiling technique with different tile sizes. Our experimental results show the tiling approach improves speed by 56.89% (2.32× faster) against straightforward (STF). And tile size of 32 has the highest speed compared to other tile sizes of 8 and 16.
Research Interests:
Texture features extraction algorithms are key functions in various image processing applications such as medical images, remote sensing, and content-based image retrieval. The most common way to extract texture features is the use of... more
Texture features extraction algorithms are key functions in various image processing applications such as medical images, remote sensing, and content-based image retrieval. The most common way to extract texture features is the use of Gray Level Co-occurrence Matrices (GLCMs). The GLCM contains the second-order statistical information of spatial relationship of the pixels of an image. Haralick texture features are extracted