The sensitive characteristic to initial value of chaos system and the immunity to noise sufficiently demonstrate the superiority in weak signal detection. In this paper Duffing equation is used as system detection model, on the basis of optimization theory, a most optimization searching method, which takes the variance of output X as the detected value is present. The basic principle and the theoretical algorithm about detecting the weak signal with this method are proposed. At the same time, the simulation experiments and the result analysis are given. The results indicated this method is rapidly, simple, convenient and the accuracy is high, which is a novel detecting frequency method. If this method were applied in signal processing field or other application field, it would have practical significance.
The main conceptual aspect of this work is on 1/f signal estimate. In this paper a new estimation algorithm that bases on Dual-Tree Complex Wavelet is proposed, which uses the variance of the wavelet coefficients at different scales to estimate the parameters of 1/f process. Adopting Maximum Posteriori estimator estimates the wavelet coefficients of 1/f process. The simulation results show that the method is effective. And comparing with other methods this method doesn't need to know the statistical characteristic of the added white noise and the parameters of the 1/f process.
This article puts forward a new measurement of departures from straightness, flatness, roundness and cylindricity in order to conquer the defect of current method on efficiency and precision of data processing, an optimized mathematical model for departure calculation is established respectively. The value of form error is calculated rapidly and precisely on microcomputer by the powerful and optimized calculation function of MATLAB. The solution of departure is established according to minimum zone. It is applied to data processing for dense measurement point especially.
New structure low intensity x-ray image system is mainly made of a plane plate mode x-ray intensifier of single proximity focus, a CCD data acquisition and a processing subsystem. This paper presents noise source inside the x-ray image system and its characteristics. By analyzing its components, image noise source of the system is found to contain quantum noise, particulate noise and dark noise of CCD. Thus a new combination method of "multi-frame mean with morphological transform filter" is studied in this paper for imaging noise elimination in the low intensity x-ray image system. Firstly, some frames of images are superimposed for mean image calculation based on the principle of noise non-correlation. Secondly, different with traditional methods of morphological transform filtering algorithm, difference image information is referred in the algorithm for source image de-noising. Based on multi-scale morphological principle, the difference image, obtained from the source image, contains both image noise and characters. After subsequent processing of wavelet translation and fuzzy algorithm, the noise of the difference image is eliminated. Thus when the processed difference image is added onto the last filtered image by the multi-scale morphological filter, the advanced image without noise is achieved which still keeps the source image characters.
Complementation image is often used as a guard technology in the trademark and paper currency. The key point of recognizing this kind of images is judging the complementary effect of complementation printing. The perspective images are usually not clear and legible, so it is difficult to recognize them. In this paper, a new method is proposed. Firstly, capture the image by reflex. Secondly, find the same norm to man-made pair printing. Lastly, judge the true and false of paper currency by the complementary effect of complementation printing. This is the purpose of inspecting the false. Theoretic analysis and simulation results reveal that the effect of man-made pair printing is good, the method has advantages such as simplicity, high calculating speed, and good robust to different RMB. The experiment results reveal that the conclusion is reasonable, and demonstrates that this approach is effective.
Three methods of waveform recovery for signals in the presence of noise are discussed in this paper. LIA (Lock-in Amplifier) is used to detect the DC signal or slowly varying signal with low SNR (signal noise ratio), the core part of which is PSD (phase sensitive detector). As a special case, the orthogonal vector LIA is discussed in detail, which is capable of vector analysis of signal. Sampling integral and digital average can recover the impulse waveform with rapid rising and roll-off edges, which have lots of harmonic components in the presence of noise. But many periods of signals are usually needed in this method. On the contrary, the correlation detection method is effective for waveform recovery of the periodical signal when signal of enough duration is not available. By simulation in Matlab, the effectiveness of digital average as well as correlation detection was verified in their specific application conditions.
The aircraft has always suffered non-stationary random vibration in the booster-phase of the launch process. Obtain the relationship between the slip-mean-time and the bias error and that between the slip-mean-time and the random error in the estimation of a non-stationary random vibration signal's power spectral density (PSD) by analyzing the estimation of the signal's variable mean square (MS). Based on this, Obtain the analytic relationship between the variable MS of a sample of a non-stationary random vibration signal and the optimum slip-mean-time which causes the smallest total error in the spectrum analysis of the signal, propose an approach for optimizing the slip-mean-time in the estimation of a non-stationary random vibration signal's PSD. Present an appliance for a signal sample of a random vibration process which is caused during an aircraft booster-phase. Used the orthogonal polynomial approximation to get the analytic format of the process's MS, confirm the optimum slip-mean-time used in estimating the signal's PSD.
In this paper, the method of empirical mode decomposition is introduced and the arithmetic of empirical mode decomposition is improved. The local mean with the envelope method of the extrema is substituted by an improved method, in which the local mean is computed by using window average method. The improved arithmetic is described in detail and the simulating model to validate the improved method is also presented in the paper. The results show the improved method has decreased the times of spline interpolation and saved the decomposition time. Moreover, all the data has participated in the computation of the local mean, thus the data to be used is increased in the improved method. Based on the improved empirical mode decomposition method, the defects of ball bearings are diagnosed practically and the results show that the proposed method is more powerful than traditional envelope analysis.
Focusing on the searching strategy in image matching, this paper constructs an energy function with features of a convex function based on Lyapunov Stability Theorem. It thus enables the Gradient neural network to converge steadily into the set of critical points of the target function. Then this paper tries to apply the network in image matching with moment invariants as the feature parameter. The specific steps of the experiments are supplied in this paper. According to the results of the experiments, this matching algorithm features good convergence, high speed, wide applicability and an extraordinary matching effect.
In vector predictive coding of speech signal sequences, the speech vector sequences that is obtained after a classification of adjacent sample of source speech sequences, can be regarded as autoregressive vector sequences with stable convariance. The vector predictive coding algorithm with highly parallel processing ability, can be achieved with the application of the principle of orthogonal projection in Hilbert space. In comparison with traditional lattice methods, the vector predictive coding algorithm has obvious advantages of calculating complexity and storage.
A novel algorithm for the detection of long-range weak targets in a sea and air scenario is proposed. A preprocessing model of weak targets in sea and sky clutter background is established based on fractal theory, which is accordant to the concept that singular local fractal dimension (LFD) will be determined at edges. The model is evaluated by comparing row-mean subtraction filter based on fractal dimension preprocessing with conventional operators such as, high pass filter and row-mean subtraction filter. Results show a similar performance in a low-noise environment and superiority of the fractal operators in a high noise and long range dim targets, the algorithms are effectively for suppression sea and sky clutter and are easy to be implemented by parallel processing hardware.
Adaptive decision feedback equalizer (ADFE) derived from transverse equalizer uses decision output signal to form a delay line, through which parts of coefficients are added together and are returned to the output. ADFE can compensate for communication channels with severe inter-symbol interference. The blur on image can be regarded as perturbation of the pixels, so in this paper a kind of ADFE is applied to 2D image restoration. There are two adaptive filers in this method, the forward filter has more flexibility to select its coefficients and is not direct inverse of blur transfer function. As a result, noise amplify will not happen. Additionally, the support of the object is determined with threshold. The experiments show that this new algorithm is robust and effectively when existing additive noise.
Partial differential equations (PDEs) are well-known due to their good processing results which it can not only smooth the noise but also preserve the edges. But the shortcomings of these processes came to being noticed by people. In some sense, PDE filter is called "cartoon model" as it produces an approximation of the input image, use the same diffusion model and parameters to process noise and signal because it can not differentiate them, therefore, the image is naturally modified toward piecewise constant functions. A new method called a directional adaptive diffusion filter is proposed in the paper, which combines PDE mode with wavelet transform. The undecimated discrete wavelet transform (UDWT) is carried out to get different frequency bands which have obviously directional selectivity and more redundancy details. Experimental results show that the proposed method provides a performance better to preserve textures, small details and global information.
The shooting range test is an important field in modern weapon development. The modern weaponry is developing towards long distance and automation directions, therefore the shooting range test is put forward new higher requirements. A novel method of target detection based on the digital image processing technology is proposed in the paper. Experiments indicate the strategy is fit to the request of the dynamic target detection and tracking in the shooting range.
Traditional spatial filtering methods such as median filtering and mean filtering are widely used to remove additive noise in images. Images smoothed using these methods are inevitably blurred. If images are fringe patterns and filtering mask used is in fixed shape, such as a rectangle of mxn pixels, the structures of the pattern processed may be even deformed. To resolve this problem, a filtering method with varying mask based on gray gradient is proposed. This method makes full use of the feature of fringe patterns in that the gradient of gray level at a pixel is normal to the tangential direction of the fringe there. According to the method, gradient vector at a pixel is calculated first, and then the shape of the filtering mask to be used is determined based on the gradient vector. If its amplitude is smaller than a predefined value, a nxn square mask is used, on the other hand, if its amplitude is larger than the predefined value, a bar mask normal to the vector is used. Finally, the median value or mean value of the pixels in the mask is then achieved as the smoothed value of the pixel. After the same operations are applied to all pixels in the pattern sequentially, the smoothed version of the fringe pattern is achieved. A simulated fringe pattern with noise is smoothed using this method and other filtering methods for comparison, and the result shows that noise in the fringe pattern can be effectively removed and the characteristic details of the fringes can be reserved at the same time.
A novel method of multi-decision of multi-focus image fusion on the basis of wavelet transform is presented for the sake of standing out clear characteristics and keeping detail information of original images. Low frequency part utilizes definition whose evaluating function is sum of Laplacian operator in the eight directions. High frequency parts utilize spatial frequency, which is calculated only one direction that is the same as high frequency direction, whose method decreases workload. Besides the traditional evaluating measures of fused image quality, which are entropy and cross entropy, the newest universal measures which combines objective and subjective factors, and the spatial frequency presented by this paper based on three directions which are horizontal, vertical and diagonal, are also used. These evaluating methods don't require a reference image. The experimental results show that the fused image obtained by proposed method has good effect, and the proposed method is superior to the other given algorithms.
In order to study on the fiber optical gyro (abbreviated as FOG) signal based on wavelet, this paper researches the FOG signal drift model and the properties of wavelet analyzed noise, introduces the wavelet filtering method, wavelet base selection, soft and hard threshold value de-noising algorithm and compulsive filtering based on The Haar wavelet. These threshold value filtering results of both of the soft and of the hard threshold value for the same wavelet base of db4 with the same Donoho threshold values and these results of compulsive filtering based on The Haar wavelet and db4 wavelet are presented also in this paper and then these main conclusions based on foregoing analysis are reached: Larger the resolving scale is, the filtering effect is more perfect. The soft threshold value filtering effect is better than that of the hard threshold value filtering at the cost of calculation when the threshold value is same. The zero shift of the compulsive filtering is least when both the wavelet and the resolving scale are same for these filtering methods. For the compulsive filtering with same wavelets, the filtering effect of Harr is better than that of db4 and the calculation of the former is fewer. Finally the author point out that applying the compulsive filtering with the Harr wavelet base and suitable resolving scale to the signal processing of FOG be helpful for the FOG's design and manufacturing.
To meet the ultra low-power and low-voltage applications,this paper presents a analog VLSI implementation of DOG wavelet transform using a balanced log-domain integrator with class AB. The circuits of implementing wavelet transform are composed of analog filter ,whose impulse response is the required wavelet . Levenbery-Marquardt nonlinear least square method is employed to obtain the transfer function of the filter.The filter design is based on IFLF structure with balanced log-domain integrators as the main building blocks.By changing the values of the log-domain integrators bias currents, this circuit can realize various scales wavelet transform. And the validity of the proposed method is confirmed by Pspice simulation results.
Using borescope equipment to inspect the inside of turbine engines is an important technology to the daily damage detection of aeronautic engine. Because the borescope image that we observe is based upon point light, and the quantum nature of light is not ideal enough, borescope image acquired through charge-coupled device (CCD) is contaminated by white Gaussian noise. Towards this, a kind of spatially adaptive context-based wavelet shrinkage borescope image denoising method was presented. The spatially adaptive wavelet thresholding was selected based on context modeling, which was used in our prior borescope image compression coder to adapt the probability. Each wavelet coefficient was modeled as a Gibbs field distribution. Context modeling was used to estimate the thresholding for each coefficient. This method was based on an overcomplete non-subsampled wavelet representation, which yielded better results than the orthogonal transform. Experimental results show that spatially adaptive wavelet thresholding yields significantly improved visual quality as well as lower mean squared error (MSE) compared to the method of Chang.
An algorithm of detection and removal for the inveracious edge of shadow in borescope images based on wavelet transform is presented. Because inveracious edges which shadows result in will influence the validity of subsequent image recognition, it is necessary to remove those inveracious edges that are extracted directly from borescope images. The grey levels of shadow regions in borescope images are low and uniform. Therefore, contours of shadows are step-structure borders. Thus, the algorithm first extracts all the edges from image using dyadic wavelet transform at multiscale. Further, the borders of shadow regions are detected from all the edges by using the property that the wavelet coefficients of step-structure borders are not correlative to the scales of wavelet transform, and connected. Finally, a neighborhood function is constructed to recognize and remove those pixels on the inveracious border. Experiment results show that this algorithm can directly obtain the intrinsical edges.
Intelligent setting system based on biomechanics and bone fracture therapy can accomplish micro-wound, intelligence and high efficiency of fracture setting. X-ray images grabbed by C-shape-arm X-ray machine supply the most key data for intelligent setting. Processing, analysis and transmission security of the image is the core in the system. According to characteristics being shown in three dimensions gray distribution figure and frequency spectrum of the image, histogram equalization in space domain and homomorphic filtering in frequency domain are separately proposed to enhance contrast and sharpness. On the foundation of mining orthopedics experts experience knowledge, setting for femoral-neck fracture is turned into three in-continuous operations that are reflected in the X-ray images through nine points, six lines, two angles and one distance and that are able to be implemented by mechanical manipulator and control device in the system. Master-slave reference frame is put forward to supply a stable reference standard to calculate parameters. Encryption method based on chaos dynamics system is brought forward to ensure image information security in the process of telemedicine intelligent setting for fracture. Clinic experience proved that the system can help orthopedists to correctly and reliably complete setting for bone fracture.
Improvement of absolute phase calibration in phase-measuring profilometry is presented. In order to calculate the absolute phase of calibration plane, linear interpolation of unwrapping phase of reference point is introduced. Accuracy of height calibration and measurement is improved. A novel 3D surface shape measurement system is designed. Experiments of given height plane are presented. The mean of measurement error of conventional algorithm is about 0.5 mm. That of the novel algorithm introduced in this paper is reduced to about 0.2 mm.
Platinic electrodes and a biologic enginery testing system are used to determine spontaneous electrical signals in the phloem, cambium and xylem of Osmanthus fragrans for the first time. The signals are denoised by the wavelet soft-threshold de-noising method, which are statistically analyzed in the time domain. De-noised signals retain the true element (the difference less than or equal to 2.6924μV) of original signals well. Results show that the plant spontaneous electrical signals are "μV" in the dimension, and the range of its amplitude in the phloem, cambium and xylem is -80.6878μV~32.3479μV, -41.1557μV~118.0153μV and 284.5316μV~393.1831μV respectively. The direction of the electrical signal in the phloem and cambium varies with the time aperiodically, while the direction of the electrical signal in the xylem is invariable. Its amplitude is significantly higher than the electrical signal in the phloem and cambium.
Heat sources recognition is very important for Printed Circuit Board (PCB) infrared thermal imaging diagnosis and several recognition techniques have been proposed by former researchers. In this paper the heat sources detection based on wavelet transforms is investigated. Cubic B-spline function is chosen as the smoothing function and its first derivative as the wavelet function. Fast recursive decomposition algorithm for 2-dimensional signal is used to compute wavelet transforms at different scales. Then the temperature gradient distributions on the thermal image of PCB under test can be determined and the bound of the heat sources can be located on the base of the gradient information. Multi-scale edge detection technique offered an opportunity to recognize the heat sources at the same time distinguishing them from the noises by choosing proper detection scale. Experimental results suggest that heat source in PCB thermal image can be recognized successfully using the proposed method.
Under high magnification of the micro-vision systems, the original metal texture appears in the form of bright spots with different shapes and different sizes and mixes with the real flaws such as maculae, pockmarks, and corrosions. The background of the image is complex while the image brightness is non-uniform due to the spherical reflection. It is difficult to extract the flaws accurately by means of traditional segmentation methods. In this paper, a new method for detecting flaws automatically is presented, in which the flaws are located approximately at first by removing the uniform background and the bright spots of metal surface texture, then detected accurately using the method of double-window. The experimental results show that this method is effective to detect various flaws on the surface of metal sphere.
According to the digital imaging of remote sensing camera and informatics theory, information entropy is presented as the automatic exposure criteria for remote sensing camera and its performance is detailedly analyzed. Aiming at the characteristics of gray-level histogram at different exposure time and the relationships between image brightness and information entropy, improved MCS (Mountain Climbing Servo) is proposed to approach the optimal exposure time in short order. The proposed algorithm has been successfully tested to be effective on a subminiature CMOS remote sensing camera system developed by the authors. Experimental results demonstrate that the proposed criteria can adjust the exposure time automatically according to the image information with high speed, sensitivity and reliability. It is of great value for CMOS camera to improve the quality of space imaging.
So far, INS/satellite integrated navigation systems have been applied widely in aviation, aircraft automatic approach and landing, land vehicle navigation and tracking, marine applications, and surveying, etc., due to their complementary characteristics. Before the data fusion, the synchronization of data from various systems is the vital challenge and significant guarantee of the integrated navigation system performance. A novel synchronization approach is presented that models the asynchronous time between INS and satellite system data and then estimates it on-line by the Kalman filter in the process of integration. Compared with other synchronization methods, this approach is simple but practical, without any increasing load of the system hardware. And simulation results indicate that the asynchronous time can be estimated real time and the estimated error is smaller than 0.1s. Meanwhile, the accuracy of the integrated navigation system is enhanced remarkably. On the ground of simulation results, it is demonstrated that the synchronization approach proposed in this article is effective.
To improve tracking performance of multi-target tracking, a data association and passive tracking scheme based on the joint probabilistic data association (JPDA) and the Gaussian sum particle filter (GSPF) is proposed. The GSPF is presented to formulate the problem of passive tracking. Compared with sampling importance resampling (SIR) scheme, GSPF can incorporate the most current observations into the particle filter and generate accurate proposal density distribution for the particle filter. Hence, the proposal scheme uses GSPF to track the states of the targets and applies the idea of JPDA directly to the sample sets of multi-target states, and weights of particles are evaluated through the combination of JPDA. The specific implementation steps of JPDA based on GSPF (JPDA/GSPF) are deduced. Trajectory tracking and the root mean square error (RMSE) comparisons are made with JPDA/SIR and JPDA/EKF schemes on simulated data in multi-target passive tracking. Simulation results show that the JPDA/GSPF scheme has better performance than JPDA/SIR and JPDA/EKF scheme in tracking. Furthermore, from the view of particle cost, the JPDA/GSPF introduces higher computation efficiency than JPDA/SIR.
Genetic algorithm is a non-derived random optimization method based on the regulations of nature selection and evolution. Generally, when genetic algorithm is applied to wavelength selection, only how to improve the prediction accuracy of a regression model is concerned. But the robustness of a regression model, that is, the anti-interference ability towards external measuring conditions variance (such as the ambient temperature, place and so on), is usually ignored. Therefore, when the measuring conditions of the predicted samples change, the regression model would predict the measured samples with high prediction errors. In this paper, genetic algorithm combined with experimental design method was studied to increase the robustness of the multivariate calibration model. In our experiments, the training set was divided into the calibration set and the monitor set to establish the regression model. The spectra of the calibration set samples were measured under the ordinary measuring conditions. The measuring conditions when obtaining the monitor set sample spectra could be arranged according to experimental design method. Kennard/Stone algorithm was used to select monitor set samples from training set. The calibration model could be built with the calibration set samples and optimized with the monitor set samples measured under the designed measuring conditions. And then the validation set samples, which were independent of the training set ones, were employed to evaluate the prediction ability of the regression model. In order to obtain a regression model with high prediction accuracy and robustness, the spectral information caused by the changes of measuring conditions need to be considered and those wavelengths which were easily interfered by external measuring factors need to be rejected when the calibration model was trained. In this paper, the modified wavelength selection method of genetic algorithm was applied to the temperature experiments of the glucose aqueous solution samples. Results revealed that not only fewer wavelengths or principal components were needed to build the calibration model but also the robustness and prediction accuracy of the calibration model were greatly improved. This modified method not only makes the regression model insensitive to external measuring conditions, but also could be applied to the calibration transfer between different instruments of the same type.
An equiripple FIR linear-phase digital filters design approach is proposed based on a novel neural network optimization technique. Its goal is to minimize the weight square-error function in the frequency domain. The design solution is presented as a parallel algorithm to approximate the desired frequency response specification, and the weight coefficients are updated according to error function. Thus, the proposed approximation method can avoid the overshoot phenomenon which may happen near the pass-band and stop-band edge of the designed filter, and may make a fast calculation of the filter's coefficients possible. Several optimal design examples are given and the performance comparison between the proposed design approach with some conventional methods, and the results show that the proposed neural network method can easily achieve higher design accuracy.
Image segmentation is a fundamental image processing technology. There are many kinds of image segmentation methods, but most of them are problem oriented. In this paper, image segmentation method based on lateral inhibition network is presented. Lateral inhibition network is a biological vision model. When an image is filtered by a lateral inhibition network, its low frequency components are inhibited while the high frequency components are enhanced. The lateral inhibited image is much easier to be segmented because of its increased inter-class difference and decreased intra-class difference. The parameters of the lateral inhibition network model determine the inhibited image, thus affect the image segmentation result greatly. But there are no assured rules to determine the parameters. We propose an evolutionary strategy (ES) based method to search the optimal weighting parameters of the lateral inhibition network model. The objective function of ES is a multiattribute fitness function that combines multiple criteria of clustering and entropy information. The original image is filtered using the optimal lateral inhibition network and then the inhibited image is segmented by an optimized threshold. Using test images of various characteristics, the proposed method is evaluated by four objective image segmentation evaluation indexes. The experimental results show its validity and universality.
After more than ten years' research efforts on the Micro Aerial Vehicle (MAV) since it was proposed in 1990s, the stable
flying platform has been matured. The next reasonable goal is to implement more practical applications for MAVs.
Equipped with a micro radio-localizer, MAVs have the ability of localizing a target that transmitting radio signals, and
further can be a novel promising Anti-Radiation device. A micro radio-localizer prototype and its localization principle
and localization algorithm are proposed. The error analysis of the algorithm is also discussed. On the basis of the
comparison of the often-used radio localization method, considering the MAVs' inherent limitation on the dimension of
the antennas, a signal intensity and guidance information based localization method is proposed. Under the assumption
that the electromagnetic wave obeys the free-space spreading model and the signal's power keeps unchanged, the
measuring equations under different target motions are established. Localization algorithm is derived. The determination
of several factors such as the number of measuring positions, numerical solving method and initial solution is discussed.
Error analysis of the localization algorithm is also proposed by utilizing error analysis theory. A radio-localizer prototype
is developed and experiment results are shown as well.
Empirical mode decomposition is a newly developed method used to analyze nonlinear and nonstationary signals. It has
been applied to many engineering domains and has represented some unique advantages. However, in the sifting process
of empirical mode decomposition, there is an involved end issue during the course of fitting the upper and lower
envelops of the signal by cubic spline function. To deal with the problem, an improved empirical mode decomposition is
proposed in this paper. It is based on combining extrapolating extrema with mirror extension. The proposed method can
solve the problem that the boundaries of the signal will swing widely when the cubic spline interpolation is used to
construct the two envelopes of the signal. The capabilities of the proposed method, the method based on mirror extension
and the method based on AR model extension algorithm will be compared in the simulation experiments. The
experimental results demonstrate that the proposed method can restrain the end effects effectively.
Echo data hiding is an important approach of audio information hiding in the time domain. But it has serious
disadvantages, such as low capacity, relatively low restoration rate and the possibility of malicious tampering, etc. In this
paper, based on the simulation analysis of the relation between the detection performance (restoration rate) and the key
parameters, we proposed a novel echo M-ary data hiding system using backward and forward echo kernel to improve the
weak points of conventional echo data hiding. The subjective quality of the stego audio and the original audio has no
evident difference and reach a good hiding effects. The capacity is increased from 50bit/s to 250 bit/s and the restoration
rate reaches to 99% for speech in 8000Hz sample frequency, 16 bits quantization, i. e., M=32. And then, we also
simulated the robustness performances, such as against adding white noise, resample, filtering, ADPCM compression,
Writer identification has become a hot topic in pattern recognition and machine learning research area. This paper studies
on the technology of text independent writer identification based on texture analysis. At first in the preprocessing stage
the uniform texture images are created from the input document. An approach for improved characters segmentation is
presented based on analysis for the character elements and their topological relations. Then the 32-channel Gabor filter is
utilized to extract 64 texture features of writing image by calculating the mean values and the standard deviations of
filtering output images. Finally, multi-class support vector machines (SVM) classifier is adopted to fulfill the
identification task. The experiment result shows that the scheme is effective and promising.
The chaotic system is sensitive to certain signals and immune to noise at the same time, so the signal-to-noise-ratio (SNR) is improved greatly when the signal embedded in noises passes through the Duffing oscillator system in great period motion state. Based on the SNR improvement property of the Duffing system, a phase estimation method of weak sine signal is proposed. The phase of the output signal is estimated firstly, the relation between the output signal phase and the input signal phase is obtained, and so the phase of the input signal can be got. Digital simulation results prove that the estimation accuracy of the present method is better than both of our previous work and traditional ML estimation method applied to input signal in noise directly. Compared with our previous work, less data is needed.
The Doppler effects caused by radar-target relative motion is one of the central problems in SF Radar, which cause
range-velocity coupling and wave distortions. Extensive research is being carried out on various velocity compensation
methods, such as varieties rate of range, Min-Entropy, PD processing and correlation methods. However, these methods
have different velocity ambiguity and velocity measurement precisions. This paper presents a new method to estimate
velocity, which can resolve the problem of ambiguous velocity by PD processing. Firstly, the waveforms of
Stepped-frequency pulse trains and PD Processing methods are analyzed. The relation between unambiguous velocity
range, precision and signal parameters is discussed. Then, a new algorithm is developed to achieve wide unambiguous
velocity and high anti-noise performance, which is similar with the principle of multiple PRFs in Pulsed Radar. Based on
parameters in some radar, the algorithm is evaluated. The simulation results confirm the above mentioned achievements.
Robust real-time tracking of non-rigid objects is a challenging task. The difficulty in visual tracking is how to match
targets from frame to frame quickly and reliably. Mean shift algorithm (MSA) is a typical nonparametric evaluation
algorithm that needs great computation. Some scholars join Kalman filter to perform state prediction in the mean shift
algorithm for reducing the computing of template matching. However, traditional Kalman filter sometimes can't track
human movement very accurately because of the particularity of human joint. While wavelet moment has the
multiresolution properties in addition to the invariant to the translation, scaling and rotation, so it is suitable for
differentiating the details of the motion objects. Therefore, Kalman-mean shift tracking algorithm based on wavelet
moment (W-K-MSA) is proposed in this paper. In this algorithm, a Kalman filter algorithm, which is used to estimate the
motion parameters of targets, is improved based on wavelet moment features in the searching process. And searching
window is adaptively changed, as a result, searching scope is reduced greatly, and the processing velocity and veracity is
improved during model matching. The experimental results demonstrate that the proposed tracking algorithm is robust
3D (three-dimensional) color digitization of an object is fulfilled by light-stripe method based on laser triangle principle
and direct capturing method based on the color photo of the object. With this system, information matching between 3D
and color sensor and data registration of different sensors are fulfilled by a sensor calibration process. The process uses
the same round filament target to calibrate all of the sensors together. The principle and procedure of the process are
presented in detail. Finally, a costume model is 3D color digitized and the obtaining data sets are processed by the
method discussed, the results verify the correctness and feasibility of the algorithm.
Fiber Optic Gyroscopes (FOGs) have been investigated and proposed as alternative sensors to magnetometers in
borehole surveying applications due to their compactness, ruggedness, low cost and high environmental insensitivity.
However, FOGs are subject to high measurement noise from various sources, which deteriorates the performance and
quality of FOGs, thus the overall system accuracy is limited. To improve the accuracy of the surveying system, adaptive
filtering techniques are utilized to reduce the noise level at the output of the FOG. A Forward Linear Prediction (FLP)
filter based on Normalized Least-Mean-Square (NLMS) adaptive algorithm was designed and evaluated using kinematic
data. Results show that the FLP filter can suppress the FOG noise to a certain degree and a satisfactory signal-to-noise
ratio improvement can be achieved using this method.
For a SIMO system, when all sub-channels have no non-zero common zeros and channel input signals have zero mean
and are temporally uncorrelated, the sufficient and essential condition of equalizing channel is that equalizer output
signals are also temporally uncorrelated. Based on this theory, this paper proposes a new equalization algorithm. The
proposed algorithm estimates directly equalizer without knowing the channel impulse response. The proposed algorithm
is based on equalizer output temporally uncorrelated characteristic and second-order statistics of the received signals. A
16QAM complex SIMO multipath channel system is simulated with the proposed algorithm in this paper. Simulation
results show that the proposed algorithm has a good equalization performance.
Feature point matching is the most common one among all kinds of stereo matching. However, since feature points are
unique, the disparity map through feature matching is also sparse. In this paper, proposed a dense disparity estimation
method that combines the reliability of feature-based correspondence methods and a reliable feature operator. The new
operator uses the principal moments of the phase congruency information to determine corner information. The resulting
corner operator is highly localized and has responses that are invariant to image contrast. This results in reliable feature
detection under varying illumination conditions with fixed thresholds. Selecting those feature points that allow left-right
correspondence based on phase correlation surrounding each point. And use the sparse correspondences at feature points
as a constraint to control the computation of dense disparity via regularized block matching that minimizes matching and
disparity smoothness errors. Experimental results show that this method can eliminate many kinds of outliers effectively.
Micro Inertia Measurement Unit based on MEMS component, is the core of the Micro Navigation System. Its accuracy
has a crucial effect on system precision. Eliminating stochastic noise in MIMU signal is of great significance to increase
the system accuracy. Aiming at the different characteristics showed at every scale space after wavelet analysis on MIMU
signal, an adaptive filtering method with decomposition level and threshold value self-adaptive adjusting is proposed by
this paper. The compactly supported Daubechies4 (db4) orthogonal wavelet is applied to decompose the signal in
multi-scale space with self-adaptive level based on white noise sequence check. An improved self-adaptive threshold
decision making is adopt for threshold filtering. After removing high frequency detail items generated by stochastic
noise, inverse wavelet transform is applied to reconstruct the original signal. The experimental results indicate that the
method can eliminate MIMU stochastic noise effectively and achieve satisfactory accuracy. And the algorithm is simple
An airborne vehicle such as a tactical missile must avoid obstacles like towers, tree branches, mountains and building
across the flight path. So the ability to detect and locate obstacles using on-board sensors is an essential step in the
autonomous navigation of aircraft low-altitude flight. This paper describes a novel method to detect and locate obstacles
using a sequence of images from a passive sensor (TV, FLIR). We model 3D scenes in the field-of-view (FOV) as a
collection of approximately planar layers that corresponds to the background and obstacles respectively. So each pixel
within a layer can have the same 2D affine motion model which depends on the relative depth of the layer. We formulate
the prior assumptions about the layers and scene within a Bayesian decision making framework which is used to
automatically determine the assignment of individual pixels to layers. Then, a generalized expectation maximization
(EM) method is used to find the MAP solution. Finally, simulation results demonstrate that this method is successful.
Template matching is one of the best and the most widely used pattern recognition method. Normalized cross-correlation
(NCC) is the main matching algorithm for template matching method. For templates with significant gray-level
variations, also called features, normalized cross-correlation can be a very simple and effective template matching
algorithm, even in cases of noisy data and changing lighting level. In the application of automatic optical inspection of
printed circuit board, many electronic components have labels on them and can be used as features for cross-correlation
template matching. However, there are quite a few components that have no labels, like some type of capacitors and
transistors. They are identified by the color or the shade of gray-level instead. These components pose great difficulties
for traditional normalized cross-correlation, which will pick up some random variations as features instead and cause
false alarms. People used to deal with this problem by including part of the background into the template to create some
artificial features, or by selecting some alternative special algorithms. Both of these methods are not ideal. Because the
first method will make the template less universal and subjects to background variations; while the second method will
loose many of the nice properties of cross-correlation algorithm. We propose an improved image cross-correlation
algorithm, which can recognize both feature based templates and uniform color or gray-level based templates. Compare
with the traditional cross-correlation algorithm, this new algorithm can be more accurate and more universal.
Experiments have shown that this new algorithm can detect feature-rich, uniformly colored, and uniformly gray
Computer vision is gaining significant importance as a cheap, passive, and information-rich sensor in research areas such
as unmanned vehicle. Using computer vision can estimate relative 3D position and attitude. This paper puts forward a
new faster relative 3D position and attitude approach based on special four feature points. This method used prior
knowledge of the four feature points of a square and parallel relation, avoided complicated iterative arithmetic in general
four feature points' methods, and reduced time of objects position and attitude estimation. The theory and experiments
with simulated data showed that the approach is efficient, robust and real-time.
Multi-class support vector machine by fusing a class of binary support vector machines is proposed. The classifier fusion approaches include simple combination method such as Maximum, Minimum, Product, Mean, Median and Major Voting. Dempster-Shafer fusion method is also presented as well as KNN and Neural network approaches. The proposed algorithms are applied to the facial expression recognition applications for both the Japanese female facial expression database and the Cohn-Kanade AU-coded facial expression database. The results show that it is effective to combine binary support vector machines (SVM) to a multi-class SVM.
Nowadays, for the BP neural network based outdoor traffic sign recognition problems, the recognition rate is generally
between 60% and 70%. Based on the results analysis, one may come to a conclusion that the key factors affecting
recognition rate are the color distortion caused by the color complexity. This paper present a new solution according to
the idea of simplifying the complex problem, using color information and intelligent approach. The first step is to break
the complex color information down to 5 kinds of standard color, and then employ BP neural network to classification.
In this article BP network is used for Color Standardization, selecting 23 normalization signs as training set and 531 real
signs as testing set for BP network. By doing so 100% average recognition rate is achieved. At the same time, it shows
the better robustness of the proposed approach for the color distortion of traffic sign in terms of either the structure
parameter or the training parameter of network.
In order to enhance the precision of videogrammetry for the profile of large-scale antenna, multi-station network is
necessary, which will also affect the reliability, efficiency and stability of the results. In this paper, the impacts of normal
and convergent configurations on the precision of space targets are discussed. Constraints, including image scale, depth
of field, field of view, and distribution of image points, which affect the network configurations, must be taken into
account. Specific network configurations for the large-scale gossamer inflatable antenna are designed and experimental
data results verify the conclusions.
The refractive index distribution in atmosphere or fluid medium is irregular and inhomogeneous, and cannot be
described by a usual gradient index formula due to the influence of flowage and change of temperature. In this paper a
novel method, namely, self-adaptive grid is used for describing the refractive index in irregular inhomogeneous medium,
the refractive index data are stored in RAM as a dynamic octree, the criterion of grid division for refractive index and its
gradient are given and the interpolation method is used for calculating the refractive index and its gradient at positions
along the ray trace trajectory. A gradient index (GRIN) rod in which the refractive index can be described by a formula is
also analysis as an example, the self-adaptive grid is created and the refractive indexes and its gradient at some positions
with high accuracy are output. The RMS error is under 7e-5 and can be used for ray tracing.
The effective way of solving integrative design, estimate and optimization of complicated multi-satellite system is
establishing integrative simulation software. This paper adopts Object-Oriented and some other interrelated technique,
setting purpose on erect flexible simulation platform which means the software should be universal, credible, flexible,
compositive and easy. It presented designing project, system structure, realization strategy, and modules of satellite
constellation simulation software system. At the same time the paper adopts the method of realizing the visualization of
satellite track under Windows platform using Visual C++ and OpenGL, which can make the software intuitionistic and
visual. Through the computation of simulation, the coverage performance and implementing cost of GEO-TDRSS and
MEO-TDRSS are analyzed and compared. The analysis and simulation results indicate that the MEO-TDRSS can
provide better coverage performance with more implementing cost.
Aimed at the application of satellite attitude control/energy storage flywheel, outer-rotor air-cored permanent magnet
brushless direct current machines (BDCM) with Halbach magnet array and the normal one are analyzed comparatively.
A prototyped BDCM with Halbach array is designed and fabricated to verify the analysis and satisfy the performance
demand of flywheel system.
Developing the satellite positioning and navigation system independently is a huge project. Establishing the simulation
and estimate software system for navigation receivers can supply reference data for developing and improving real
receivers. This paper analyzed the general structure of the receiver, presented a more feasible design method for the
receiver simulation and estimate software system, introduced some estimate items and the data or algorithms which the
items needed, and proposed a solution to C/A code rapid acquisition with inertial navigation system (INS) assisting. This
software system consists of a signal simulator, a navigation receiver simulator and a receiver estimate software and is
based on MATLAB/SIMULINK. It helps to configure a precise receiver simulation and estimate system.
This paper provides an introduction to Wireless Sensor Networks (WSN), their applications in the field of control
engineering and elsewhere and gives pointers to future research needs. WSN are collections of stand-alone devices
which, typically, have one or more sensors (e.g. temperature, light level), some limited processing capability and a
wireless interface allowing communication with a base station. As they are usually battery powered, the biggest
challenge is to achieve the necessary monitoring whilst using the least amount of power.
In common with most disciplines surface metrology is having to evolve in order to meet the new requirements demanded
by miniaturization and the increased use of semi conductor planar technology. Changes are taking place in theory as well
as in the other constituents of an engineering design such as material properties and new manufacturing processes. In
this paper a number of issues will be discussed. They will be by no means a comprehensive list but sufficient will be
investigated to give an idea of the nature and scope on the problem. Amongst the topics covered will be nomenclature,
some mechanical properties, force balance etc. There will be a look back at the way in which surface metrology evolved
to see if lessons can be learned.