Mirjana Bonković

Population-Based Uncalibrated Visual Servoing

This paper introduces the implementation of a recently introduced method suitable for visual servoing. The method is based on the generalization of secant methods for nonlinear optimization. The difference with existing approaches related to visual servoing is that we do not impose a linear model to interpolate the goal function. Instead, we prefer to identify the linear model by building the secant model using population of the previous iterates, which is as close as possible to the nonlinear function, in the least-squares sense. The new system has been shown to be less sensitive to noise and exhibits a faster convergence than do conventional quasi-Newton methods. The theoretical results are verified experimentally and also by simulations.

Two-level coarse-to-fine classification algorithm for asthma wheezing recognition in children's respiratory sounds

Abstract The paper proposes a two-layer pattern recognition system architecture for asthma wheezing detection in recorded childrens respiratory sounds. The first layer consists of two SVM classifiers specifically designed as a cascade stacked in parallel to emphasize the differences among signals with similar acoustic properties, such as wheezes and inspiratory stridors. The second layer is realized using a digital detection threshold, which further upgrades the proposed structure with the aim of improving the process of wheezing detection. The results were experimentally evaluated on the data acquired from the General Hospital of Dubrovnik, Croatia. Classification results obtained on the test data sets revealed that the central frequency of wheezes included in the training data is important for the success of classification.

New Kinematic Parameters for Quantifying Irregularities in the Human and Humanoid Robot Gait

Gait patterns of humans and humanoid robots are often described by analysing changes in angular rotation of hip, knee and ankle joints during one gait cycle. Each joint displays specific behaviour and irregularities of the gait pattern could be detected by measuring displacements from the normal rotation curve, while small deviations of individual gait characteristics are usually not easily detected. In this paper, an advanced gait analysis method is proposed, which incorporates analysis of angular data and its derivations of hip, knee, and ankle joints, presented in the phase plane. The gait kinematics was measured using a system based on active markers and fast digital cameras. The experiment included measurements on thirty healthy, barefoot humans while walking on a treadmill. We also simulated types of irregular gait, by measurements on subjects wearing knee constraints. The new kinematic parameters which are introduced clearly indicated the discrepancy between normal, healthy gait trials and irregular gait trials. The proposed gait factor parameter is a valuable measure for the detection of irregularities in gait patterns of humans and humanoid robots.

Solving a two-colour problem by applying probabilistic approach to a full-colour multi-frame image super-resolution

Reconstruction based algorithms play an important role in the multi-frame super-resolution problem. A group of images of the same scene are fused together to produce an image with higher spatial resolution, or with more visible details in the high spatial frequency features. Demosaicing algorithms interpolate missing pixels in a raw image taken from one Charged Coupled Device (CCD) array, upsampling the number of the pixels present in the image. Since super-resolution (SR) and demosaicing are the two faces of the same problem it is natural to address them together. In this paper it is: (i) shown that correct modelling of the Bayer pattern in the generative process improves the super-resolution performance for colour images, and (ii) an algorithm that incorporates the two colour prior into the probabilistic model is designed. The algorithm presented in this paper focuses on the classes of images that have two dominant colours, i.e. most of the areas in the image are uniformly coloured. A convex optimization procedure for joint super-resolution and demosaicing is developed which outperforms state-of-the-art algorithms.

Sleep EEG Composition in the First Three Months of Life in Monozygotic and Dizygotic Twins

We investigated genetic influence on sleep electroencephalogram (EEG) composition by a classical twin study of monozygotic (MZ) and dizygotic (DZ) twins in the first 3 months of life. Polysomnographic (PSG) recordings were obtained in 10 MZ and 20 DZ twin pairs in the 37th, 46th, and 52nd week of postmenstrual age (PMA). The EEG power spectra were generated on the basis of fast Fourier transformation (FFT). Genetic influence on active sleep/rapid eye movement (AS/REM)] and quiet sleep/ non rapid eye movement (QS/NREM) sleep composition was estimated by calculating within pair concordance and the intraclass correlation coefficients (ICCs) for delta (0.5-3.5 Hz), theta (4-7.5 Hz), alpha (8-11.5 Hz), sigma (12-14 Hz), and beta (14.5-20 Hz) at central derivation. MZ twins show higher ICCs than DZ twins for alpha, sigma, and beta spectral powers during QS/NREM sleep in the 37th, 46th, and 52nd week PMA. However, there was no significant difference (P > .05) between the 2 types of twins in absolute differences of EEG spectral power of the alpha, beta, and sigma frequency ranges in the 37th, 46th, and 52nd week PMA. The greatest mean absolute difference within MZ and DZ twin pairs and also between MZ and DZ twin groups was identified in the delta frequency range. Our findings gave an indication of genetic influence on alpha, sigma, and beta frequency ranges in the QS/NREM sleep stage.

Comparison of Uncalibrated Model-Free Visual Servoing Methods for Small-Amplitude Movements: A Simulation Study

The paper compares the performance of several methods used for the estimation of an image Jacobian matrix in uncalibrated model-free visual servoing. This was achieved for an eye-in-hand configuration with small-amplitude movements with several sets of system parameters. The tested methods included the Broyden algorithm, Kalman and particle filters as well as the recently proposed population-based algorithm. The algorithms were tested in a simulation environment (Peter Corkes Robotic Toolbox for MATLAB) on a PUMA 560 robot. Several application scenarios were considered, including static point and dynamic trajectory tracking, with several characteristic shapes and three different speeds. Based on the obtained results, conclusions were drawn about the strengths and weaknesses of each method both for a particular setup and in general. Algorithm-switching was introduced and explored, since it might be expected to improve overall robot tracking performance with respect to the desired trajectory. Finally, possible future research directions are suggested.

Local colour statistics for edge definition

This paper addresses important computer vision problem of defining a measure for the strength (intensity) of edges in an image. The importance of the problem arises from the fact that images are typically represented in the multi - colour RGB format, hence each pixel requires more than one parameter to be defined. The α image is created using local colour statistics based on the size of the neighbourhood, surrounding image pixel under consideration. The local colour statistics defined by the size of the neighbourhood surrounding is called patch. The paper introduces measure for definition and detection of edges in the image based on that local colour neighbourhood. The valuation measure is based on the sum of absolute measure present in α image. This way, the number of parameters required for defining pixels and edges is reduced to a single parameter, α. Therefore, patch size plays an important role in definition and detection of edges.

Application of EEMD-ICA algorithm to EMG signals measured in laryngeal muscles

This paper describes the application of EEMD-ICA algorithms on electromyographic signals measured in laryngeal muscles. The method was used for the separation of singlechannel data into independent components. During the speech, there was a transcranial magnetic stimulation of the motor cortex area of the brain for speech production i.e. primary motor region of the laryngeal muscles (M1) and Brocas region. Manifestation of magnetic stimulation of those cortex areas and speech itself is recorded in the form of electromyographic signals in laryngeal muscles. The measured signals are a mixture of two different sources: natural stimulus (speech) and the effect of electromagnetic stimulation depending on the area of the speech cortex that is stimulated. This research demonstrated that using EEMD-ICA method, signal which is a mixture of speech and the effect of electromagnetic stimulation to specific areas of the speech cortex, can be successfully separated to the original components. The results were obtained using Matlab. The impact of magnetic stimulation to brain regions is detected and isolated from the laryngeal muscle signal.

Modeling and analysis of the impact of nTMS on the EMG signal of laryngeal muscles

This paper describes a methodology for the analysis of electromyographic signals. Biosignal measurements were performed with the application of electromagnetic stimulation of specific regions of the cerebral cortex and the response was recorded from electrodes hooked to laryngeal muscles. Analysis of the resulting signal is needed to identify markers in the measured signal relating to electromagnetic stimulus and thus provide background for mapping brain regions characteristic for the production of speech. Given the fact that the neurosurgical priority is to preserve these regions (eloquent or critical regions of the cortex) of injury during surgery because such injury can cause permanent damage, it is necessary to pay great attention to operation planning. This requires high-quality mapping. Methods of electrical stimulation produce results, but are applied intraoperatively. The aim is to develop a non-invasive method and for this purpose electromagnetic stimulation is more appropriate. The paper describes the model of stimulus propagation and acquisition as well as signal analysis using methods of principal components analysis for blind source separation and spectrogram analysis filtered with with wavelet packets. Software tool MATLAB is used for signal processing.

Evaluation of fuzzy clustering methods for segmentation of environmental images

The article compares three methods for segmentation of environmental images. Hue and saturation values of the image pixels were used as the input values for the clustering. The methods that have been examined are K-medoid, fuzzy Cmeans and Gustafson-Kessel. Results of the fuzzy clustering methods were compared with the results obtained with method using the mean-shift algorithm.