Sharif Digital Repository

Steganography is method for communicating confidential information through a non-trustworth in way which hides the existence of communication. For improving the security of steganography statistical detectability must decrease as such as possible. Despite the fact, that the quality of the relation between statistical detectability and amount of distortion engendered by embedding is still an open problem, problem of detectability reduces to problem of management of pixel embedding in order to minimization of distortion. As in wet paper coding methods, an optimum (or approximately optimum) algorithm proportioned to Pixel-cost has been offered, the current problem of steganography is to find... 

In recent years, Intelligent Transportation Systems (ITS) have received special attentions both in research and in commercial areas. Increased infrastructure facilities, like surveillance cameras, has made this concept even more attainable than before. In this respect, the ability to automatically extract information from traffic images, as one of the key inputs of ITSs, is of great importance. With an increased number of surveillance cameras and the need for more accurate information regarding the road users and their interactions, in order to better city traffic management, building and repairing roads, trip time estimation, number of people per roads estimation and etc, using human... 

With the advent of technology, small in size sensors, memory, speeding up the processor and lowering the cost, it is possible to build an embedded camera system. The goal of this project is to design and build an embedded camera system so it can execute any set of necessary algorithms as depending on the application. In this project, two models of embedded camera systems have been presented as an integrated system and a system with independent units. To design the integrated embedded system, ZYNQ processor is used and two structures are presented in the form of hardware-software and hardware design. In hardware-software design, image processing operations are done by software and in hardware... 

Smartphone sensors like accelerometer, gyroscope and magnetometer are very common nowadays. This gives us the opportunity for sensor-based activity recognition. This thesis's goal is to collect data from different smartphone sensors and then extract hand-crafted features and classify them using machine learning algorithms. Metro, bus, taxi, bicycle, running, upstairs, walking and standing are studied activities in this thesis. All above steps are covered in this research, later we want to present an activity recognition model and then test it through a web server, after that, we modify the model by proposing to change learning coefficient to gain better accuracy. Finally, an Android app was... 

Predicting traffic condition is one of the most important topics for traffic engineers and is a vital part of smart city. Extracting traffic behavior and realizing new traffic behaviors plays undeniable parts in big decision making of cities. On the other hand, progress of new technologies and devices for controlling and monitoring traffic caused the volume of traffic related data to grow exponentially. Velocity and variety of these data made normal data mining solution to fail in modeling and performing well. New data era has changed the view of large numbers of science and engineering majors. Traffic engineering also needs big data analytics for process of its own big data. Thus the need... 

The multi people tracking is considered a fundamental problem in computer vision, which has received considerable attention from academic and commercial fields. This issue deals with a set of proposed methods that track the movement path of several humans in a video-like sequence. The problem of multi people tracking is the foundation of other computer vision problems, including human gesture estimation, motion recognition, and behavioral analysis, and is mainly used in emerging fields such as automatic car driving, smart security, service robots, etc. Although many methods have been proposed and investigated to solve the above problem; But there are still serious challenges, such as severe... 

Traffic density estimation is one of the most challenging problems in Intelligent Transportation Systems. One of the important traffic information that is broadcasted to drivers is Traffic Density information. In many traffic control centers; human operators are responsible for estimating traffic density from captured video data. Increasing traffic cameras and constraint number of operators introduce an updating delay to broadcasted information. So it is important to have an automatic traffic density estimation system. In this thesis, machine vision is used to solve this problem. Supervised Image classification is our approach. In supervised Image classification, images are classified to... 

Statistical detectability is a feature of a steganalyzer that states its power for distinguishing between cover and stego images. A steganographer tries to design the steganography algorithm such that the steganalyzer could not detect the stego images. As a result, designing an efficient steganography algorithm (assigning the pixel cost) that could minimize the statistical detectability is an important objective in steganography. However, making a sensible relation between the pixel cost and statistical detectability is an open problem.
In this thesis, by modeling the steganalyzer using the LDA topic model, we analyze its error probability as a criterion of statistical detectability.... 

Audio signal processing is a greatly useful approach to the Internet of Things since analyzing prominent audio signals can provide valuable information about environmental activities. Environmental sound processing is used in applications such as mechanical systems diagnosis, industrial maintenance, security systems, etc. This approach requires the design and development of sound classification and detection systems. In this thesis, we have achieved 84.5% accuracy on optimizing the features (by feature engineering and feature learning) and exploiting different types of machine learning algorithms. Well-known databases such as ESC-50 have been used to test and evaluate the whole system. Among... 

Nowadays, using different types of data has shown significant impacts on analyzing the related systems. Growth in data volume, systems complexity and existence of error and obscurity in collecting the data, increased the necessity of inventing new data analysis methods. Location-based data is an important data type for such analyses which are collected from sensors in different places. These data besides other official organization's information like municipality or Google … provide us a bulk volume of raw data. Such collections of raw data are mostly diverse, heterogeneous, bulk and outspread. Inspite of that, raw data with machine learning algorithms lead to considerable practical... 

In the transportation industry, one of the most important and fundamental problems is the traffic of vehicles in the transportation roads. This problem is especially seen in large and densely populated cities such as Tehran. If traffic control is not done properly, it can lead to problems such as reduced traffic dynamics, environmental pollution, wasted drivers' time, disorder and loss of energy. If the traffic control is done after creating a traffic problem, it will not bring good results and will have low efficiency. For this reason, optimal traffic management and control has been raised as an important issue, especially in large cities. Predicting traffic flow is one of the important and... 

The comfort of occupants has been a key driver behind the adoption of smart building technologies in recent years. To achieve this goal, machine learning methods have seen significant growth, with deep reinforcement learning emerging as particularly advantageous. Unlike advanced control strategies such as model predictive control, deep reinforcement learning does not require a physical model. Instead, learning occurs through the direct interaction of an intelligent agent with the environment, without the need for predefined datasets. This approach is further strengthened by its ability to adapt to dynamic conditions and effectively operate in large state spaces. Given that heating and... 

Environment Identification and recognizing surrounding objects is an exigent need in future applications. For example one of the emerging technologies in car industry is driverless cars. In driverless cars, navigation system should be able to detect and recognize pedestrians, traffic signs, roads, surrounding cars and so on. Therefore, conventional single-object recognition systems are not capable of handling the needs of advanced machine vision based applications. In recent years, designing and analyzing multi-class object detection and recognition systems have become a big challenge in machine vision. In this thesis our goal is to identify and analyze the existing problems in designing... 

Increasing needs for traffic video surveillance and intelligent video analysis is a hot topic that has attracted significant interests in recent days. Scene understanding is one of the most important aspects of an intelligent video surveillance system. Statistics shows that rate of vehicle accidents has been continuously increasing over the past years. This necessitates the need for efficient traffic analysis and abnormality detection systems. It is desirable to develop fully automated surveillance systems, with which people will be free from boring monitoring tasks. In this thesis our goal is to describe and detect abnormal event like accident or variation of traffic density in traffic... 

Detection and Tracking objects in various environmental conditions is a challenging task in computer vision. Abrupt changes in illumination, object size and noise level make this task even harder. Due to these problems, it is somehow impossible to propose a fully functional system for tracking every type of object. In this thesis we propose a new method for detection and tracking vehicles in traffic scenes. The problem is solved by dividing it into two parts: at First we have evaluated more than ten state-of-the art trackers, the final multi-vehicle tracker is chosen, and then, like any other object tracking method, we have used a unique identiy for each vehicle. This traffic identity is... 

Motion pattern analysis in traffic videos can be directly employed for scene analysis, rule mining, abnormal event detection, traffic phase detection, etc. The most successful and newest methods for complex traffic scene analysis are based on topic models. Topic models have been developed for text processing and as they have not been yet optimized for traffic video processing, they are still far away from optimal efficiency. In this thesis, firstly we propose two unsupervised methods for traffic video analysis based on non-probabilistic topic models. In the first proposed method, we use Group Sparse Topical Coding (GSTC) and an improved version of it for learning typical motion patterns... 

Data analysis is the science of extracting human-understandable knowledge from data. Massive data analysis is usually associated with the challenges of Volume, Velocity, Variety, Veracity and Value. These challenges are often referred to as 5V. With increasing processing power and reducing its cost in the last decade, massive data analysis methods have been used in various fields such as market, insurance and health, telecommunications and network, web search engines, intelligent transportation, etc. In this dissertation, we have tried to extract applied knowledge by using massive data analysis platforms and algorithms as well as traffic data. In this dissertation, a mathematical model for... 

Emotion recognition from speech signals has become one of the most popular researches in recent years. In order to increase human-machine interaction, a proper connection must be established between them. To achieve this goal, a machine must be able to understand the situation and respond accordingly. Part of this process involves understanding the user's emotional state. In recent years, various methods have been proposed to increase the efficiency of the speech emotion recognition system. These methods include collecting various audio databases, extracting efficient features from speech signals, using feature selection algorithms, designing different classifiers, as well as combining... 

Blind image steganalysis is a technique used for detecting the existence of the data hidden in an image, where no information about the stenographic algorithm is available or usable. In this way, an important problem is to find sensitive features which make noticeable statistical distinction between cover and stego images. New steganalysis methods based on multi-resolution transform, specifically the wavelet and the contourlet transforms, have been proposed in this thesis in order to enhance the detection accuracy of system especially at low embedding rates. In fact, multi-resolution transforms are powerful space-frequency analysis tools that have been found quite successful in detection of... 

Widespread transfer of digital files over networks provides a hidden channel to transfer secret messages. Current steganalysis schemes need to work on a complete image for doing the detection job that starts when the image is entirely transferred, so are often restricted to an offline process. This restriction is serious when existence of the hidden message carriers on the network is shorter than the time required for the detection process. In this thesis, we propose a structurally fast detection method to detect the data hidden in an image passing through network. We use two of most powerful steganalysis algorithms for steganalysis of images that proposed by 1) Fridrich and Pevny and 2) Liu...