Sharif Digital Repository

With the extensive growth in information, text classification as one of the text mining methods, plays a vital role in organizing and management information. Most text classification methods represent a documents collection as a Bag of Words (BOW) model and then use the histogram of words as the classification features. But in this way, the number of features is very large; therefore performing text classification faces serious computational cost problems. Moreover, the BOW representation is unable to recognize semantic relations between words. Recently, topic-model approaches have been successfully applied for text classification to overcome the problems of BOW. Our main goal in this paper... 

Analyzing motion patterns in traffic videos can directly generate some high-level descriptions of the video content which can be further employed in rule mining and abnormal event detection. The most recent and successful unsupervised approaches for complex traffic scene analysis are based on topic models. However, most existing topic models share some key characteristics which could limit their utility. In this paper, based on extracted optical flow features from video clips, we employ Sparse Topical Coding (STC) framework to automatically discover typical motion patterns in traffic scenes. For this purpose, we improve the STC to overcome one of the drawbacks of topic models with the aim of... 

In visual surveillance, detecting and localising abnormal events are of great interest. In this study, an unsupervised method is proposed to automatically discover abnormal events occurring in traffic videos. For learning typical motion patterns occurring in such videos, a group sparse topical coding (GSTC) framework and an improved version of it are applied to optical flow features extracted from video clips. Then a very simple and efficient algorithm is proposed for GSTC. It is shown that discovered motion patterns can be employed directly in detecting abnormal events. A variety of abnormality metrics based on the resulting sparse codes for detection of abnormality are investigated.... 

Traffic density estimation plays an integral role in intelligent transportation systems (ITS), using which provides important information for signal control and effective traffic management. In this paper, we present a new framework for traffic density estimation based on topic model, which is an unsupervised model. This framework uses a set of visual features without any need to individual vehicle detection and tracking, and discovers the motion patterns automatically in traffic scenes by using topic model. Then, likelihood value allocated to each video clip enables us to estimate its traffic density. Results on a standard dataset show high classification performance of our proposed... 

This paper addresses the optimality of Network topology for localization purposes when measurements are range-only. Localization procedures based on TOA and TDOA range measurements are considered in this paper. As of now, the optimal placement of fixed terminals to locate mobile terminals has been studied based on Cramer-Rao Lower Bound (CRLB). CRLB is a universal bound on the variance of a general unbiased estimator and therefore is not dependent to the estimator and methods of localization. Here we introduce another algorithm dependent criterion, based on perturbation theory of linear equations. Different optimal topologies are presented using the new criterion. These optimal topologies,... 

Traffic density estimation is one of the most challenging problems in Intelligent Transportation Systems. In this paper, we estimate the traffic flow density based on classification. Various new efficient features are introduced for distinguishing between different traffic states, including number of key-points, edges of difference-image and moving edges. These features describe the traffic flow without any need to individual vehicles detection and tracking. We experiment our proposed approach on a standard database and some real videos from Tehran roads. The results show high accuracy performance of our method, even in changes of environmental conditions (e.g., lighting), by using efficient... 

A two-level Sparse Topical Coding (STC) topic model is proposed in this paper for analyzing video sequences of traffic surveillance containing hierarchical patterns accompanied by complicated motions and co-occurrences. In order to automatically cluster optical flow features into motion patterns, a first level STC model is used. Next, the second level STC model is applied for clustering motion patterns into traffic phases. The effectiveness of the suggested method is proved by experiments on a traffic dataset in the real world. Our simulations show that the proposed two-level STC is able to extract the motion patterns and traffic phases accurately, leading to realistic describing the traffic... 

Reflection appears as a layer that partly covers the original image. This phenomenon may cause failure in extracting information from images of reflecting objects. This work presents an automated technique for determining of reflection area as well as its severity to define reliability of extracted information. This is done by analyzing histogram and objects found in the image. We present a strong detector based on combining the results of these two procedures. We presented new reflection detection algorithm and a new method to find a good threshold value for converting any image into a binary image. The average reflection detection accuracy of the proposed algorithm is more than 95% for... 

A new steganalysis system for JPG-based image data hiding is proposed in this paper. We use features extracted from both wavelet and DCT domains that are refined later in the sense of utmost discrimination between the clear and stego images in the classification system. Statistical properties of the SVD of wavelet sub-bands are combined with the extended DCT-Markov features, and the features that are most sensitive to the data embedding are chosen through a SVM-RFE based selection algorithm. Experimental results show significant improvement over baseline methods, especially for steganalysis of Perturbed Quantization (PQ), which is known to be one of most secure JPG-based steganography... 

In this paper, the topic model is adopted to learn traffic phases from video sequence. Phase detection is applied to determine where a video clip is in the traffic light sequence. Each video clip is labeled by a certain traffic phase, based on which, videos are segmented clip by clip. Using topic models, without any prior knowledge of the traffic rules, activities are detected as distributions over quantized optical flow vectors. Then, traffic phases are discovered as clusters over activities according to the traffic signals. We employ the Fully Sparse Topic Model (FSTM) as the topic model. The results show that our method can successfully discover both activities and traffic phases which... 

Analyzing motion patterns in traffic videos can directly lead to generate some high-level descriptions of the video content. In this paper, an unsupervised method is proposed to automatically discover motion patterns occurring in traffic video scenes. For this purpose, based on optical flow features extracted from video clips, an improved Non-negative Matrix Factorization (NMF) framework is applied for learning of semantic motion patterns. After extracting the motion patterns, each video clip can be sparsely represented as a weighted sum of learned patterns which can further be employed in very large range of applications. Experimental results show that our proposed approach finds accurately... 

In this work, we present a novel FPGA-based implementation of the AES algorithm which has a two-layered resistance against power analysis attacks. Our countermeasure is based on the concept of finite state machine equipped with a random number generator. Beyond masking the intermediate variables as the first layer of defense, we randomize the sequences of operations and add dummy computations as the second layer of defense. Therefore, the first order attack is prevented and the number of power traces needed for a successful second order attack is vastly increased and the correlation coefficient is decreased, as expected. © 2017 IEEE  

The advent of automated vehicles (AVs) and the sharing of these vehicles in traffic flow has raised researchers' interest in low-cost solutions to traffic congestion, especially at intersections as points of traffic flow interlace. Also, as AVs have not been used commercially yet, it is impossible to analyze the effects of their sharing on traffic flow. However, different scenarios of AVs alongside human-driven vehicles (HDVs) can be modeled and simulated. By defining three levels of automated vehicles along with HDVs, this paper investigates the impact of AVs on intersection capacity by applying a series of hypotheses to the fundamental traffic flow formula. Different scenarios with an AV... 

In this paper the effect of emittance on the relevant issues of the synchrotron radiation sources, optics and beamline design for two values of emittance, ex=3.278 and 0.937 nm.rad, suggested for the Iranian Light Source Facility lattice design, have been discussed. The effect of spot size and divergence of the electron beam, and strength of the bending magnet field, have been considered in the calculations. The results show that reducing of emittance, increases the brilliance of the undulator by factor 5, and decrease the foot print on the optical elements and spot size of photon beam on the sample by a factor of about 2  

Many high performance steganographic schemes work at a limited or sparsely distributed set of embedding rates. We have shown that some steganographic changes will be wasted as these schemes are utilized individually for messages of various lengths. To measure the wasted changes and compare different schemes in this respect, we have built a framework based on two new criteria: the Relative Change Waste (RCW) and the Expected Changes per Pixel (ECP). To decrease the wasted changes a systematic combination of schemes is introduced and proved to be equivalent to nonlinear interpolation of points in a two-dimensional space. We have proved that a special case which leads to a linear interpolation... 

A family of matrix embedding steganographic schemes for digital images is investigated. The target schemes are applied to blocks of n pixels in a cover image. In every block, at most m pixels are allowed to change with q-ary steps. We have derived some upper bounds on the embedding efficiency of these schemes for different values on m. It is also shown that these upper bounds approach the general upper bound on the embedding efficiency of q-ary steganography. For the case of q=3, we have shown that there is no feasible optimum member of the family for m=2, although for m=1, a well-known example exists. Instead, for m=2, a new close-to-bound scheme in the family is presented which exploits... 

A family of matrix embedding steganographic schemes for digital images is investigated. The target schemes are applied to blocks of n pixels in a cover image. In every block, at most m pixels are allowed to change with q-ary steps. We have derived some upper bounds on the embedding efficiency of these schemes for different values on m. It is also shown that these upper bounds approach the general upper bound on the embedding efficiency of q-ary steganography. For the case of q=3, we have shown that there is no feasible optimum member of the family for m=2, although for m=1, a well-known example exists. Instead, for m=2, a new close-to-bound scheme in the family is presented which exploits... 

In this paper, a new formulation for discriminative training of HMMs is introduced as a solution to several speech recognition problems. This formulation uses a properly trained MLP in a simple interconnection with HMMs called "Cascade HMM/ANN Hybrid". The training algorithm has simple realization in comparison with other discriminative training for HMMs such as MDI and MMI. Also a rigid mathematical proof of its convergence has been presented. No significant increase in computational requirements is needed in recognition phase and the recognition task can still be performed in real-time. This structure has been employed in some isolated and continuous speaker-independent speech recognition...