2007/1
Browse
Recent Submissions
- ItemThe Effects of MIMO Antenna System Parameters and Carrier Frequency on Active Control Suppression of EM Fields(Společnost pro radioelektronické inženýrství, 2007-04) Mohammed, Abbas; Hult, TommyIn this paper we propose a new approach employing adaptive active control algorithms combined with a Multiple-Input Multiple-Output (MIMO) antenna system to suppress the electromagnetic field at a certain volume in space (e.g., at the human head). We will investigate the effects of the size and number of MIMO antenna elements on the system performance and test the algorithms at different carrier frequencies (e.g., GSM bands and UMTS).
- ItemFace Recognition Methods Based on Feedforward Neural Networks, Principal Component Analysis and Self-Organizing Map(Společnost pro radioelektronické inženýrství, 2007-04) Oravec, Milos; Pavlovicova, JarmilaIn this contribution, human face as biometric is considered. Original method of feature extraction from image data is introduced using MLP (multilayer perceptron) and PCA (principal component analysis). This method is used in human face recognition system and results are compared to face recognition system using PCA directly, to a system with direct classification of input images by MLP and RBF (radial basis function) networks, and to a system using MLP as a feature extractor and MLP and RBF networks in the role of classifier. Also a two-stage method for face recognition is presented, in which Kohonen self-organizing map is used as a feature extractor. MLP and RBF network are used as classifiers. In order to obtain deeper insight into presented methods, also visualizations of internal representation of input data obtained by neural networks are presented.
- ItemAnimation of 3D Model of Human Head(Společnost pro radioelektronické inženýrství, 2007-04) Mihalik, Jan; Michalcin, ViktorThe paper deals with the new algorithm of animation of 3D model of the human head in combination with its global motion. The designed algorithm is very fast and with low calculation requirements, because it does not need the synthesis of the input videosequence for estimation of the animation parameters as well as the parameters of global motion. The used 3D model Candide generates different expressions using its animation units which are controlled by the animation parameters. These ones are estimated on the basis of optical flow without the need of extracting of the feature points in the frames of the input videosequence because they are given by the selected vertices of the animation units of the calibrated 3D model Candide. The established multiple iterations inside the designed animation algorithm of 3D model of the human head between two successive frames significantly improved its accuracy above all for the large motion.
- ItemNeural Networks for Synthesis and Optimization of Antenna Arrays(Společnost pro radioelektronické inženýrství, 2007-04) Merad, Loo; Bendimerad, Fethi Tarik; Meriah, Sidi Mohamed; Djennas, Sidi AhmedThis paper describes a usual application of back-propagation neural networks for synthesis and optimization of antenna array. The neural network is able to model and to optimize the antennas arrays, by acting on radioelectric or geometric parameters and by taking into account predetermined general criteria. The neural network allows not only establishing important analytical equations for the optimization step, but also a great flexibility between the system parameters in input and output. This step of optimization becomes then possible due to the explicit relation given by the neural network. According to different formulations of the synthesis problem such as acting on the feed law (amplitude and/or phase) and/or space position of the radiating sources, results on antennas arrays synthesis and optimization by neural networks are presented and discussed. However ANN is able to generate very fast the results of synthesis comparing to other approaches.
- ItemImplementations of HVS Models in Digital Image Watermarking(Společnost pro radioelektronické inženýrství, 2007-04) Foris, Peter; Levicky, DusanIn the paper two possible implementations of Human Visual System (HVS) models in digital watermarking of still images are presented. The first method performs watermark embedding in transform domain of Discrete Cosine Transform (DCT) and the second method is based on Discrete Wavelet Transform (DWT). Both methods use HVS models to select perceptually significant transform coefficients and at the same time to determine the bounds of modification of selected coefficients in watermark embedding process. The HVS models in DCT and DWT domains consist of three parts which exploit various properties of human eye. The first part is the HVS model in DCT (DWT) domain based on three basic properties of human vision: frequency sensitivity, luminance sensitivity and masking effects. The second part is the HVS model based on Region of Interest (ROI). It is composed of contrast thresholds as a function of spatial frequency and eye\'s eccentricity. The third part is the HVS model based on noise visibility in an image and is described by so called Noise Visibility Function (NVF). Watermark detection is performed without use of original image and watermarks have a form of real number sequences with normal distribution zero mean and unit variance. The robustness of presented perceptual watermarking methods against various types of attacks is also briefly discussed.
- «
- 1 (current)
- 2
- 3
- »