Sharif Digital Repository

This thesis proposes new algorithms for digital watermarking that are based on Independent Component Analysis (ICA) technique. First, we will show that ICA allows the maximization of the information content and minimization of the induced distortion by decomposing the covertext (in this thesis the image) into statistically independent components. In fact, for a broad class of attacks and fixed capacity values, one can show that distortion is minimized when the message is embedded in statistically independent components. Information theoretical analysis also shows that the information hiding capacity of statistically independent components is maximal. Then we will propose a new wavelet... 

Structural system identification from vibration data is one of the most interesting research topics in the structural health monitoring area. Recently, realization and detection of the effects of damage when a structure is subjected to strong ground motion has become a great concern in earthquake and structural engineering communities. Seismic signal processing is one of the most reliable methods of detecting the structural damage during earthquakes. The structural responses during earthquakes are nonstationary with respect to both amplitude and frequency. The state-of-the-art time-frequency distributions when applied to vibration records were studied. Different methods of analysis for... 

The quantity and quality of information extracted from the brain, in addition to data collection methods, is also related to the statistical tools used. As extracting maximum information in both temporal and spatial dimension require electrophysiological approaches on the physical side, the statistical methods should be optimized to that end, on the theoretical side. The time histogram method is the most basic tool for capturing a time-dependent rate of neuronal spikes. Generally, in the neurophysiological literature, the bin size that critically determines the goodness of the fit of the time histogram to the underlying spike rate has been subjectively selected by individual researchers.... 

It is generally accepted that Electronic Warfare has three distinct components: (1) electronic support (ES), (2) electronic attack (EA), and (3) electronic protect (EP). ES is included those measures taken to collect information about an adversary by intercepting radiated emissions. EA refers to attempting to deny adversaries access to their information by radiating energy into their receivers. EP includes activities under taken to prevent an adversary from successfully conducting ES or EA on friendly forces.
The function of Electronic Support Measurement (ESM) System is considered as a part of the first component (i.e. ES). After receiving emitted signals from various radars by ESM... 

Blind source separation methods such as independent component analysis (ICA) and second order blind identification (SOBI) have shown considerable potential in the area of ambient vibration system identification. The objective of these methods is to separate the modal responses, or sources, from the measured output responses, without the knowledge of excitation. Several frequency domain and time domain methods have been proposed and successfully implemented in the literature. Whereas frequency-domain methods pose several challenges typical of dealing with signals in the frequency-domain, popular time domain methods such as NExT/ERA and SSI pose limitations in dealing with noise, low sensor... 

Epilepsy is a neurological disorder whose prevalence is estimated to be 1% of the world population. Electroencephalogram (EEG) is one of the best and convenient non-invasive tools used in diagnosis and analysis of this disease. Epileptic components extracted from EEG recordings are widely used in neuroscience in the diagnosis analysis like epilepsy source localization. However, epileptic components are often contaminated and covered with artifacts of physiological origin (baseline, EMG, ECG, EOG, etc.) or instrument noises (power supply, electrode, etc.). So, preprocessing and denoising is necessary for precise analysis of epilepsy EEG recording. Heretofore, several methods have been... 

Blind Source Separation (BSS) is a technique for estimating individual source components from their mixtures at multiple sensors, where the mixing model is unknown. Although it has been mathematically shown that for linear mixtures, under mild conditions, mutually independent sources can be reconstructed up to accepted ambiguities, there is not such theoretical basis for general nonlinear models. This is why there are relatively few resultsin the literature in this regard in the recent decades, which are focused on specific structured nonlinearities.In the present study, the problem is tackled using a novel approach utilizing temporal information of the signals. The original idea followed in... 

Blind Source Separation (BSS) is a challenging task in signal processing which aims to separate sources from their mixtures when no information is available about the sources or the mixing system. Different approaches have already been proposed for source separation.However, during the last decade, new approaches based on multimodal nature of phenomena have been proposed for source separation. Different aspects of a multimodal phenomenon can be measured by means of different instruments where each of the measured signals is called a modality of that phenomenon. Although the modalities are different signals with different features, due to the same physical origin, they usually have some... 

Graph signal separation is a new topic in the field of graph signal processing that aims to recover graph signals from their linear combinations, taking into account the relationship between the signals and their corresponding graphs. Among the existing methods for separating graph signals from observing only one mixture, a recently published approach assumes the smoothness of the signals and minimizes the smoothness criterion of the signals on their related graphs. In this thesis, the closed-form solution of this method is obtained and the reconstruction error of the graph signals is calculated from it and the performance of this method is evaluated. It is also shown by numerical... 

Blind separation of mixed graph signals is one of the new topics in the field of graph signal processing. However, similar to the most proposed methods for separating traditional signals, it is assumed that the number of observed signals is equal to or greater than the number of sources. In this thesis, we show that a signal can be uniquely decomposed into the summation of a set of smooth graph signals, up to the indeterminacy of their DC values. From the blind source separation point of view, this is like the separation of a set of graph signals from a single mixture, contrary to traditional blind source separation in which at least two observed mixtures are required. Moreover, we... 

In this work, we evaluate differernt tensor decomposition methods in application of fECG extraction from abdominal ECG recordings. After selecting proper tensor decomposition tool (Tucker decomposition) we propose a linear source separation algorithm based on a measure of quasi-periodicity. The quasi-periodicity is attained through the use of a constraint on a matrix factorization problem. In practice, we form a three dimensional ”tensor” by stacking the observation matrix and rough estimates obtained by both linear and non-linear subspace reconstruction methods. The method is applied to a database of electrocardiography (ECG) recordings, where rough subspace estimates of maternal and fetal... 

The problem of separation of audio sources is one of the oldest issues raised in the field of audio processing, which has been studied for more than half a century. The main focus of recent research in this field has been on improving the sound quality resulting from the separation of sound sources with the help of deep neural networks. This is despite the fact that in most applications of audio source separation, such as the application of meeting transcription, we do not need the separated audio of people. Rather, we need a pipeline of converting overlapping speech to text, which, by receiving the audio in which several people have spoken, outputs the text spoken by the people present in... 

Cardiovascular (CV) system is very similar to a wireless communication system in which a common input signal from the heart is fed into different arterial channels throughout different body parts. By putting multiple sensors on different peripheral body sites effects of this circulation from the heart can be recorded and be used as inputs for different multi channel blind system identification (BSI) methods for estimation of arterial channel dynamics. This Thesis is defined in order to investigate different BSI methods capability in CV characterization. To achieve this goal photoplethysmogram signals has been used as primary sensory recorded effect of heart function at three different... 

The most popular system for quantification of the facial behaviors and expressions is the Facial Action Coding System (FACS). FACS provides a description of all possible and visually detectable facial variations in terms of 33 Action Units (AUs). The activation of each AU leads to a slight variation in the facial appearance, and any facial expression can be modeled by a single AU or a combination of AUs. Definition of AUs is such that they are sparse in multiple domains. The goal of this dissertation is utilizing these sparsity properties to develop an effective algorithm for automatic intensity estimation of AUs. One of the sparsity domains of AUs is the spatial domain that means the... 

Nowadays, Electroencephalogram (EEG) is the most common method for brain activity measurement. Event Related Potentials (ERP) which are recorded through EEG, have many applications. Detecting ERP signals is an important task since their amplitudes are quite small compared to the background EEG. The usual way to address this problem is to repeat the process of EEG recording several times and use the average signal. Though averaging can be helpful, there is a need for more complicated filtering. Blind source separation methods are frequently used for ERP denoising. These methods don’t use prior information for extracting sources and their use is limited to 2D problems only. To address these... 

Functional Magnetic Resonance Imaging (fMRI) is one of the imaging techniques that are used to study human brain function and neurological disease diagnosis. Popular techniques in fMRI utilize the blood oxygenation level dependent (BOLD) contrast, which is based on the differing magnetic properties of oxygenated (diamagnetic) and deoxygenated (paramagnetic) blood. In order to analyze fMRI data, hypothesis-driven or data-driven methods can be used. Among data-driven techniques, Independent Component Analysis (ICA) provides a powerful method for the exploratory analysis of fMRI data. In this thesis, we use ICA on fMRI data for detecting active regions in brain, without a-priori knowledge of... 

Analyzing cardiac function of the fetus during pregnancy is proved to be an important prenatal care procedure. Traditional methods like auscultation and ultrasonography could only lead to anatomical information about the fetal heart. So in the recent decades many researches on the abdominal electrical signals of the pregnant women have been done. Nowadays, it is possible to record the heart magnetic signals. With regard to the morphological similarity between the electrical and magnetical signals of the heart and the superiority of the magnetic ones, one could assume more diagnostic capacity for the fetal MCG. It should be mentioned that finding the location of the fetal R waves could help...