The recognition of 3-D objects is quite a difficult task for computer vision systems. This paper presents a new object framework, which utilizes densely sampled grids with different resolutions to represent the local information of the input image. A Markov random field model is then created to model the geometric distribution of the object key nodes. Flexible matching, which aims to find the accurate correspondence map between the key points of two images, is performed by combining the local similarities and the geometric relations together using the highest confidence first method. Afterwards, a global similarity is calculated for object recognition. Experimental results on Coil-100 object database, which consists of 7 200 images of 100 objects, are presented. When the numbers of templates vary from 4, 8, 18 to 36 for each object, and the remaining images compose the test sets, the object recognition rates are 95.75 %, 99.30 %, 100.0 % and 100.0 %, respectively. The excellent recognition performance is much better than those of the other cited references, which indicates that our approach is well-suited for appearance-based object recognition.

In recent years, the eigenvoice approach has proven to be an efficient method for rapid speaker adaptation, which directs the adaptation according to the analysis of full speaker vector space. In this article, we developed a new algorithm for eigenspace-based adaptation restricting eigenvoices in clustered subspaces, and maximum likelihood (ML) criterion was replaced with maximum aposteriori (MAP) criterion for better parameter estimation. Experiments show that even with one sentence adaptation data this algorithm would result in 6.45 % error ratio reduction relatively, which overcomes the instability of maximum likelihood linear regression (MLLR) with limited data and is much faster than traditional MAP method. This algorithm is not highly-dependent on subspace number of division, thus it proved to be a robust adaptation algorithm.

A new method for edge detection based on directional space is proposed. The principle is that: firstly, the directional differential space is set up in which the ridge edge pixels and valley edge pixels are abstracted with the help of the method of logical judgments along the direction of differential function, forming a directional roof edge map; secondly, step edge pixels are abstracted between the neighboring directional ridge edge and directional valley edge along the direction of differential function; finally, the ridge edge map, valley edge map and step edge map gained along different directions are combined into corresponding ridge edge map, valley edge map and step edge map. This method is different from classical algorithms in which the gray differential values of the mutual vertical direction are combined into one gradient value. The experiment of edge detection is made for the images of nature scenery, human body and accumulative raw material, whose result is compared with the one of classical algorithms and showing the robustness of the proposed method.

The traditional Snake algorithm cannot effectively detect the object edge of an image with non-convex shapes or low SNR. This paper studies the characteristics of this type of image with complex shape target or noise and presents an improved Snake algorithm. The traditional Snake function model and operation strategy are improved by increasing new control energy functions, and the influencing weight of these energy factors is discussed. At the same time, a dynamic arrangement for the Snake points is used to adapt different target shapes. The simulation results indicate that the new Snake model greatly decreases the dependence on the Snake point s initial position and effectively overcomes noise influence. This method enhances the Snake algorithm s ability of detecting object edge.

In this paper, an image edge detection method based on multi-fractal spectrum analysis is presented. The coarse grain H?der exponent of the image pixels is first computed. then, its multi-fractal spectrum is estimated by the kernel estimation method. Finally, the image edge detection is done by means of different multi-fractal spectrum values. Simulation results show that this method is efficient and has better locality compared with the traditional edge detection methods such as the Sobel method.

The traditional snake initial contour should be close to the true boundary of an object of interest in an image; otherwise, an incorrect result will be obtained. Next, active contours have difficulties progressing into boundary concavities. Moreover, the traditional snake as well as almost all of its improved methods can be easily obtained from the local minimum because snake models are nonconvex. An active contour model based on force field analysis (FFA), namely, FFA snake model, is presented in this paper. Based on analyzing force distribution rules of the distance potential force field, a standard is introduced here to distinguish the false one from contour points. The result is not considered as the final solution when the snake energy is minimal. Furthermore, estimation and calculation should be made according to the established standard; only then can the result be considered final. Thus, the snake is prevented from running into the local minimum. The simulation results show that the FFA snake model has a large capture range and can move a snake into the boundary concavities, and that it is able to obtain the object of interest s contour precisely. Compared with the gradient vector flow snake, this new model has a low computational cost.

An adaptive multi-user detector was developed for direct-sequence code division multiple access (DS-CDMA) systems corrupted by non-Gaussian channel noise, which can be quite detrimental to the performance of the multi-user detectors based on classical white Gaussian assumption. This receiver simultaneously combats multiple-access interference (MAI) and non-Gaussian impulsive noise. The channel parameters are estimated and transmitted signals are jointly detected by a simple recursive algorithm derived from the EM/SAGE algorithm. Analytical and simulation results show that the proposed technique is robust with wider applicability than conventional multi-user detectors in terms of near-far resistance and bit-error ratio (BER) when either MAI or non-Gaussian impulsive noise is dominant.

The features of low earth orbit/medium earth orbit (LEO/MEO) satellite networks routing algorithm based on inter-satellite link are analyzed and the similarities between satellite networks and mobile Ad Hoc network (MANET) are pointed out. The similar parts in MANET routing protocol are used in the satellite network for reference. A new dynamic routing algorithm based on MANET in LEO/MEO satellite networks, which fits for the LEO/MEO satellite communication system, is proposed. At the same time, the model of the algorithm is simulated and features are analyzed. It is shown that the algorithm has strong adaptability. It can give the network high autonomy, perfect function, low system overhead and great compatibility.

Due to the aspect sensitivity of high-resolution range profile (HRRP), traditional radar HRRP target recognition methods usually use average profile within some target-aspect region as the target-aspect template. Actually, the amplitude fluctuation property of target HRRP also represents some feature information of the target. Based on the scattering center model, a new feature extraction method using the amplitude fluctuation property of target HRRP is proposed in this paper. The weighted HRRP feature extracted by the new method can represent the scatterer distribution in every range cell, thereby it can describe the scattering property of the target better. The experimental results based on measured data show that the new feature extraction method can greatly improve recognition performances.

Quantitative attributes are partitioned into several fuzzy sets by using fuzzy c-means algorithm. Fuzzy c-means algorithm can embody the actual distribution of the data, and fuzzy sets can soften the partition boundary. Then, we improve the search technology of apriori algorithm and present the algorithm for mining fuzzy association rules. As the database size becomes larger and larger, a better way is to mine fuzzy association rules in parallel. In the parallel mining algorithm, quantitative attributes are partitioned into several fuzzy sets by using parallel fuzzy c-means algorithm. Boolean parallel algorithm is improved to discover frequent fuzzy attribute set, and the fuzzy association rules with at least a minimum confidence are generated on all processors. The experiment results implemented on the distributed linked PC/workstation show that the parallel mining algorithm has fine scaleup, sizeup and speedup. Last, we discuss the application of fuzzy association rules in the classification. The example shows that the accuracy of classification systems of the fuzzy association rules is better than that of the two popular classification methods: C4.5 and CBA.

This paper studied the bifurcation and chaos phenomenon in a multi-parallel-connected current-mode controlled boost DC DC converter system with the use of nonlinear mapping bifurcation theory of two dimensions, and the changing rules of the bifurcation charts with the increase of the control parallels and control parameters were concluded. The method of discrete mapping modeling was utilized to construct the difference equations of the system operating in continuous conduction mode (CCM). Analyses and computer emulations were made.

This paper presents a new method for rotor broken-bar fault diagnosis of induction motors. The asymmetry of the rotor caused by broken-bar fault will give rise to the appearance of additional frequency component of 2 sf_s (s is slip and f_s is supply frequency) in the electromagnetic torque spectrum. The startup electromagnetic torque signal is decomposed into several intrinsic mode function (IMF) with empirical mode decomposition (EMD) based on the Hilbert-Huang Transform. Then, using the instantaneous frequency extraction principle of the Hilbert Transform, the rotor broken-bar fault characteristic frequency of 2sf_s can be exactly extracted from the IMF component, which includes the rotor fault information. Moreover, the magnitude of the IMF which includes the rotor fault information can also give the number of rotor broken bars. Experimental results demonstrate that the proposed electromagnetic torque-based fault diagnosis method is feasible.

The brushless DC motor (BLDCM) with trapezoidal electromotive force (back-EMF) waveform is used widely. In principle, when the motor runs in the 120° conduction mode, two of the three phases are active while the other phase is inactive at all times. However, a ripple current occurs in the inactive phase due to the diode freewheeling during the non-commutation period in the traditional pulse width modulation (PWM) methods, which aggravates the torque ripples. A new PWM method is proposed in this paper to eliminate the diode freewheeling during the non-commutation period in the inactive phase. As a result, the torque ripple is suppressed using the proposed method. The simulation and experimental results are demonstrated to verify the validity of the proposed PWM method.

In this paper, the authors propose a new space vector pulse width modulation (SVPWM) algorithm based on non-orthogonal coordinates for N-level inverters. First, it is pointed out that classical αβ coordinates-based SVPWM has many shortcomings because of improper coordinate choice. Then, a non-orthogonal coordinates-based SVPWM is proposed to solve these problems. The proposed algorithm can easily identify which sector the reference space vector falls in and conduct simple operations to find the duty cycle of each vector. Finally, it is verified that the proposed SVPWM is actually a pulse-width modulation (PWM) technology based on line voltages.

An adjacent table-based simplified model of distribution networks containing medium voltage buses of a substation is established. Identification of bus outage and the condition to start fast restoration procedure are discussed. A complex load shading parameter is set up to describe various load shading schemes. The imaginary part of the load shading parameter describes the states of switches of load shading schemes while the real part is the corresponding amount of shaded load. A new concept of independent tripping operation is also put forward. The procedure to search the operation with the least amount of shaded load for a feeder and a connected domain are detailed. The procedure for fast restoration of a large area breakdown of the whole distribution network under emergency states is dealt with using a typical grid distribution network as an example. Results of analysis show that the direct load shading scheme under the most balanced topology is not always the optimal scheme. The proposed method can obtain the optimal operating mode with the least amount of shaded load thus showing its feasibility.

There are usually no on-line product quality measurements in batch and semi-batch processes, which make the process control task very difficult. In this paper, a model for predicting the end-product quality from the available on-line process variables at the early stage of a batch is developed using partial least squares (PLS) method. Furthermore, some available mid-course quality measurements are used to rectify the final prediction results. To deal with the problem that the process may change with time, recursive PLS (RPLS) algorithm is used to update the model based on the new batch data and the old model parameters after each batch. An application to a simulated batch MMA polymerization process demonstrates the effectiveness of the proposed method.

A modular approach of the estimation-based design in adaptive linear control systems has been extended to the adaptive robust control of strict-feedback stochastic nonlinear systems with additive standard Wiener noises and constant unknown parameters. By using Itô s differentiation rule, nonlinear damping and adaptive Backstepping procedure, the input-to-state stable controller of global stabilization in probability is developed, which guarantees that system states are bounded and the system has a robust stabilization. According to Swapping technique, we develop two filters and convert dynamic parametric models into static ones to which the gradient update law is designed. Transient performance of the system is estimated by the norm of error. Results of simulation show the effectiveness of the control algorithms. The modular design, which has a concise hierarchy, is more flexible and versatile than a Lyapunov-based algorithm.

A novel estimation scheme based on dead reckoning (DR) model for networked control system (NCS) is proposed in this paper. Both the detailed DR estimation algorithm and the stability analysis of the system are given. By using DR estimation of the state, the effect of communication delays is overcome. This makes a controller designed without considering delays still applicable in NCS. Moreover, the scheme can effectively solve the problem of data packet loss or timeout.

In the book (Adaptive Identification, Prediction and Control Multi Level Recursive Approach), the concept of dynamical linearization of nonlinear systems has been presented. This dynamical linearization is formal only, not a real linearization. From the linearization procedure, we can find a new approach of system identification, which is on-line real-time modeling and real-time feedback control correction. The modeling and real-time feedback control have been integrated in the identification approach, with the parameter adaptation model being abandoned. The structure adaptation of control systems has been achieved, which avoids the complex modeling steps. The objective of this paper is to introduce the approach of integrated modeling and control.

A white-light interferometric fiber-optic sensing network based on the double-ring topology is demonstrated, which can be applied to the measurements of quasi-distributed strain and temperature in a smart structure. In order to increase the multiplexing capacity, decrease the measurement cost of each sensor, and improve the ability of reliability of the sensor network, a double-port interrogating technology was used. The double- ring fiber optical sensing network based on the space division multiplexing (SDM) is further developed. The low coherent multiplexing principle in the double-ring network structure is analyzed. Based on the optical path matching condition of SDM, the intensity characteristic of the interference signal in the sensor is deduced. The characteristics of the double-ring sensing network connecting 9 sensors and its property of robust resisting destruction are verified by experiments, and the results are analyzed and discussed.

Deep submicron process technology is widely being used and interconnect structures are becoming more and more complex. This means that the resistance calculation based on two-dimensional models can no longer provide sufficiently accurate results. This paper presents a three-dimensional resistance calculation method called the combined analytical formula and boundary element method (ABEM). The method cuts selected interconnecting lines then it calculates the resistances of straight sections using an analytical formula and the resistances of the other sections using the boundary element method (BEM). The resistances of the different sub-regions are combined to calculate the resistance of the entire region. Experiments on actual layouts show that compared with the commercial software Raphael based on finite difference method, the proposed method is 2 3 orders of magnitude faster. The ABEM method uses much less memory (about 0.1% 1%), and is more accurate than Raphael with default mesh partitions. The results illustrate that the proposed method is efficient and accurate.

In this paper, the near-field vector components were used to combine the FEM method and the uniform-geometrical theory of diffraction (UTD) method for analyzing phased array antenna mounted on an airborne platform. First, HFSS, a set of software based on finite element method (FEM), was utilized to find the near-field vector components of the phased array antennas, and then these vector components were used as the source of the UTD method to get the disturbed radiation pattern. Numerical results show that the hybrid of the two methods not only extends the use of the UTD program, but also effectively solves this type of challenging problems.

For the finite-difference time domain (FDTD) method, the electromagnetic scattering problem, which requires the characteristic structure size to be much smaller than the wavelength of the exciting source, is still a challenge. To circumvent this difficulty, this paper presents a novel hybrid numerical technique of combined difference and spectrum for time-domain Maxwell s equations. With periodical continuation of each time-dependent quantity in Maxwell s equations, the solutions before and after the continuation remain consistent in the first period, which results in the conversion of the continuous spectrum problem to a discrete one. The discrete spectrum of the field after continuation is obtained from difference methods for Maxwell s curl equations in frequency-domain, and the time domain solution of the original problem is derived from their inverse Fourier transform. Due to its unconditional stability, the proposed scheme excels FDTD in resolving the aforementioned problems. In addition, this method can simulate dispersive media whose electric susceptibility cannot be expressed with Debye or Lorentz types of models. In dealing with boundary conditions, it can utilize the perfectly matched layer (PML) without extra codes. Numerical experiments demonstrate its effectiveness, easy implementation and high precision.