Catálogo de publicaciones - libros

It is a difficult and important task to classify the types of military vehicles using the acoustic and seismic signals generated by military vehicles. For improving the classification accuracy, we investigate different feature extraction methods and 4 classifiers. Short Time Fourier transform (STFT) is employed for feature extraction from the primary acoustic and seismic signals. Independent component analysis (ICA) and principal component analysis (PCA) are used to extract features further for dimension reduction of feature vector. Four different classifiers including decision tree (C4.5), K-nearest neighbor (KNN), probabilistic neural network (PNN) and support vector machine (SVM) are utilized for classification. The classification results indicate the performance of SVM surpasses those of C4.5, KNN, and PNN. The experiments demonstrate ICA and PCA are effective methods for feature dimension reduction. The results showed the classification accuracies of classifiers with PCA were superior to those of classifiers with ICA. From the perspective of signal source, the classification accuracies of classifiers using acoustic signals are averagely higher 15% than those of classifiers using seismic signals.

The extended Kalman Filter (EKF) is an important method for eliminating stochastic errors of dynamic position in the Global Positioning System (GPS). One of the adaptive methods is called the adaptive fading Kalman filter (AFKF), which employs suboptimal fading factors for solving the divergence problem in an EKF. Incorporation of a scaling factor has been proposed for tuning the fading factors so as to improve the tracking capability. A novel scheme called the fuzzy adaptive fading Kalman filter (FAFKF) is proposed. In the FAFKF, the fuzzy logic reasoning system is incorporated into the AFKF. By monitoring the degree of divergence (DOD) parameters based on the innovation information, the fuzzy logic adaptive system (FLAS) is designed for dynamically adjusting the scaling factor according to the change in vehicle dynamics. GPS navigation processing using the FAFKF will be simulated to validate the effectiveness of the proposed strategy.

The triangle packing problem has yielded many significant theories and applications such as textile cutting and container stuffing. Although the representation method of the popular linear quadtree has many merits, it puts too much emphasis upon the symmetry of image segmentation. Therefore, it is not the optimal representation method. In this paper, inspired by the concept of the triangle packing problem, we present a Triangle Non-symmetry and Anti-packing pattern representation Model (TNAM). Also, we propose a novel algorithm for the TNAM of the gray images. By comparing the algorithm for the TNAM with that for the linear quadtree, the theoretical and experimental results show that the former is much more effective than the latter and is a better method to represent the gray images. The algorithm for the TNAM of the gray images is valuable for the theoretical research and potential business foreground.

PDE (Partial differential equation) is an image interpolation method which interpolates based on local geometry property. It can not preserve texture pattern and can only process natural image. NL (Non Local)-means is an interpolation method that uses global information of image. Entire texture pattern in image can be well preserved because of the high replication property of NL-means, while the problem is that blur is preserved as well. In this paper a novel image interpolation method which combines PDE and NL-means is proposed. Image interpolated by the novel method is clear and smooth, and preserves texture pattern. The new method enhances edges using shock filter PDE which does not strengthen jaggies of block contour in interpolated image; the PDE used in this method to smooth image diffuses along level curve. Divided gray regions caused by PDE are smoothed by NL-means; the broken texture pattern is recovered well. Lastly, it is proved that even noisy image can be directly interpolated to the required size using this method. Both theoretical analysis and experiments have been used to verify the benefits of the novel interpolation method.

It presents an approach to the automatic generation of pencil sketching with the effects of paper texture. First, proposes a texture mapping of strokes in the aspect of skeleton, the filter algorithm and the standard deviation algorithm for rendering image, the near distances recover algorithm for real-time browsing, and finally implement the Pen-and-Ink Style of pencil sketching Rendering System. In the static rendering, it needn’t adjust the threshold for the convenient on rendering and the outlines are more precise and exact. Besides, attaching the graftal and sketching shadow makes the composition of an image more attractive. Through a series of clever image processes, the system finally presents excellent colored pencil style drawings. Because the proposed algorithm is not complicated, the rendering time is quite short compared to other past related studies. Compared with other research works and Photoshop on a set of benchmarks, the system demonstrates its strength in the aspects of full automation, stability of sketching quality and higher visual satisfaction, all achieved in a considerably shorter time.

An antenna array composed of one transmitter and multi-receivers and dedicated to measuring terminal trajectory of the target of interest, which is supposed to be in uniform rectilinear motion, is set up. On the basis of the model, the Vector parameter that can uniquely determine the terminal trajectory of target is introduced, and the measurement equations which describe the respective relationships between the Vector parameter and the instantaneous Doppler frequency and the phase differences extracted from echoes are established. Taking advantage of the measurement equations, we propose an algorithm of estimating the Vector parameter without resolving the phase difference ambiguity; furthermore, the detailed steps in estimating the Vector parameter using numerical optimization techniques are put forward. The Monte Carlo simulation results demonstrate the effectiveness and reliability of our numerical algorithm compared with the traditional method.

An inter-scale adaptive, data-driven threshold for image denoising via wavelet soft-thresholding is proposed. To get the optimal threshold, a Bayesian estimator is applied to the wavelet coefficients. The threshold is based on the accurate modeling of the distribution of wavelet coefficients using generalized Gaussian distribution (GGD), and the near exponential prior of the wavelet coefficients across scales. The new approach outperforms BayesShrink because it captures the statistical inter-scale property of wavelet coefficients, and is more adaptive to the data of each subband. Simulation results show that higher peak-signal-to-noise ratio can be obtained as compared to other thresholding methods for image denoising.

This paper describes our work in implementing a real-time moving object detection system basing on a system on a chip (SoC) system. We have implemented the algorithms necessary for moving object detection while using a SoC IP and have prepared an exclusive image-processing SoC system to implement the algorithms. The implemented IP is the IP for the image, I2C control, edge detection, TFT-LCD, median filter, SRAM control, and moving object detection. Detection of a moving object, as for the input image, requires processing edge detection, image differentiation, application of a median filter, and last, detecting the moving object. The moving object area for a detected movement detects the moving object by the cumulative value of binary conversion density.

In this paper, a strapdown inertial navigation system (SINS) error model is introduced, and the model observability is analyzed. Due to the weak observability of SINS error model, the azimuth error can not be estimated quickly by Kalman filter. To reduce the initial alignment time, a neural network method for the initial alignment of SINS on stationary base is presented. In the method, the neural network is trained based on the data preprocessed by a Kalman filter. To smooth the neural network output data, a filter is implemented when the trained neural network is adopted as a state observer in the initial alignment. Computer simulation results illustrate that the neural network method can reduce the time of initial alignment greatly, and the estimation errors of misalignment angles are within a satisfied range.

Traditional speech recognition systems have relied on power spectral densities, Mel-frequency cepstral, linear prediction coding and formant analysis. This paper introduces two novel input feature sets and their extraction methods for intelligent phoneme identification. These input sets are based on intrinsic phonetic characteristics of Arabic speech comprising of the dimensionally reduced Power Spectral Densities (DPSD) and Location, Trend, Gradient (LTG) values of the captured speech signal spectrum. These characteristics have been subsequently utilized as inputs to four different neural network based recognition classifiers. The classifiers have been tested for twenty-eight Arabic phonemes utterances from over one hundred non-native speakers. The results obtained using the proposed feature sets have been compared and it has been observed that LTG based input feature set provides an average phoneme identification accuracy of 86% as compared to 70% obtained through applying DPSD based inputs for similar classifiers. It is worthwhile to note that the methods proposed in this paper are generic and are equally applicable to other regional languages such as Persian and Urdu.