2012 Vol. 37, No. 3
The Bursa-Wolf model,molodensky model etc.are widely used in geodesy,photogrammetry and other subjects to solve the transformation parameters between three-dimension coordinates.These models are limited to the coordinate transformation of small rotation angle.But,in practice,a great many coordinate transformation of big rotation angle are needed to be determined.Traditional 3D coordinate transformation model is improved that can solve any rotation angle's 3D coordinate transformation.The examples indicate that the proposed algorithm is feasible and can determine the transformation parameters quickly.
The third frequency carrier phase virtual observation is produced by numerical transform through the physical relationship of each frequency carrier phase observations.Using long-wave character of multi-frequency combination,a group of long-wave combinations are formed to detect cycle slip of GPS double-frequency raw carrier phase observations in different conditions through the pseudorange/carrier phase combination method and get a group of cycle slip candidate.Then we use Melbourne-Wubbena combination get the difference on each epoch of wide-lane to judge the true cycle slip.The result shows that this method can real-time fix the cycle slip of double-frequency origin carrier phase observations.
In GPS/INS integrated navigation,the number of observations is usually less than that of the state parameters,and the single adaptive factor is usually applied in Kalman filtering,which can lead to precision loss of indirect observational parameters.A new algorithm of classified adaptive filtering is presented based on predicted residuals and selecting weight filtering,and the corresponding formulas are given.Finally,an actual calculation is given.The new algorithm can not only degrade the influence of the disturbances from the state but also avoid the loss of estimated precision of indirect observational parameters,and improve the accuracy of the navigation further.
Three points central differences method(CDM) has been widely used in GPS velocity and acceleration determination.However,it will lead to a considerable error when the carrier has a strong maneuverability.We prove that the truncation errors will be decreased and observational errors will be increased when more points be used in CDM.For validation of our analysis,some airborne GPS data are used.The results show that if data interval is 1 Hz,average velocity and acceleration of carrier is 20 m·s-1,0.2 m·s-2 respectively,9 points CDM is the best.Comparing of common using 3 points CDM,the precision of velocity determination is improved about 40%.And the improved level is more significant when the carrier has a strong maneuverability.
The algorithm of integrity with different frequencies of the EGNOS system is deduced.The regional integrity of EGNOS system is simulated under GPS and GPS/Galileo systems with different frequencies.The results show that the EGNOS can provide Cat-Ⅰ integrity requirement in case of GPS/GALILEO systems with double-frequency.
The two-level distributed optimal fusion algorithm which can be used in multi-sensor integrated navigation system is studied,and the theoretical mode is approved.The final fusion results with feedback and without feedback information are equivalent and optimal.The simulated results verify the above theory,and also illustrate that the feedback information from the fusion centre can obviously increase the filtering performance of every sub-filter.
Errors of gyro free inertial navigation system(GF-INS) using 6 accelerometers grow faster with time comparing with the conventional inertial navigation system(INS),for the reason that it has one more integral operation when calculating the angular velocity.To improve the accuracy of GF-INS,a novel scheme of 9 accelerometers GF-INS is proposed and a new calculative method is discussed,which can eliminate the iterative errors of integral operation for angular velocity calculation and calculate linear acceleration with the difference of outputs of accelerometers only.So the constant biases of Accelerometers and coupling errors from angular movement can been eliminated.This scheme can calculate squares of angular velocity,angular accelerations and linear accelerations with only the linear combination of outputs of 9 Accelerometers,so the computation is very simple and timesaving.This schome regards three-dimensional squares of angular velocity,angular accelerations and linear acclerations as 9 independent variables,so the 9 accelerometers GF-INS has the least accelerometers.The simulation results show that the divergence of errors is restrained remarkably with the new calculative method.
Reference datum of global crustal horizontal motion was studied.ITRF2005VEL was obtained based on Euler law and ITRF2005 velocity field,the total angular momentum of the tectonic plates was also calculated and discussed.The absolute plate motion model NNR-ITFR2005 was established based on NNR(no-net-rotation) reference frame through constraining the velocities of ITRF2005 by using the condition of discrete Tisserand,and was compared and analyzed from geological model NNR-NUVEL1A and the other ITRF series NNR models.The results show that there is a better consistency of NNR-ITRF2005 and NNR-NUVEL1A model in whole.While,at the same time,there inevitably exist some differences between NNR-ITRF2005 and NNR-NUVEL1A which are related to the factors of lacking of observation stations,uneven station distributions,short observation times,not enough constraints to plates,and some problems of themselves and so on.At last,the net rotation of the lithosphere was also discussed.
According to the time variant characteristic of Chandler wobble(CW) in LS model,we put forward a new CLS model based on correction of time variant parameters of the CW.We use the CLS+AR model to predict polar motion.The model-building is discussed in detail,and the prediction accuracy of CLS+AR model is compared with LS+AR model and other models.The results show significant improvement of the prediction accuracy of the CLS+AR model in comparison with the LS+AR model,and this proves the effectiveness and the superiority of the new model.
The relationship between polar gaps problem and three types of gravity gradiometry boundary value problem is discussed from qualitative and quantitative point of view.The simulation result shows that the zonal items of the spherical harmonic analysis solution of vertical-vertical gradiometry boundary value problem in the sphere,the one-order items of the spherical harmonic analysis solution of vertical-horizontal gradiometry boundary value problem and the two-order items of the solution of horizontal-horizontal boundary value problem are influenced mostly by polar gaps problem respectively.
Mathematical models of integral equation approach to recover the Earth's gravitational field have been presented.Several gravitational field models have been recovered based on GOCE orbits of 61 days from 2009-11-02 to 2010-01-02.The results show that the long wavelength information of gravitational field can be effectively extracted from GOCE orbits,which complements the band-limitation of GOCE gradiometers.The geoid height error is ±9.6 cm with degree and order 106.The precision is higher than that from EIGEN-CHAMP03S and GRACE model from 2 months orbits.But the precision of zonal coefficients is low due to the polar gap.The geoid error of the model which combines the orbits of GOCE and GRACE satellites is ±6.9 cm with degree and order 106,which complements the orbital deficiency of GOCE satellite.
Many methods can be used to determine smoothing parameters in semiparametric model,but smoothing parameters vary with these methods for the same model.After comparative analysis of the L-Curve,GCV(Generalize Cross Validation) and virtual observational approach,the factors influencing smoothing parameter are studied by simulated example.Smoothing parameters have a close relation with signal-to-noise ratio.When the signal-to-noise ratio increases,variation trend of smoothing parameters is different.The best method determining smoothing parameters in different situations is different.
We study the deformation of Xianshuihe Fault using CR-InSAR.Firstly,we process 6 ASAR data,which were achieved after the corner reflector(CR) installation by GAMMA software.Then we get 3 CR points(Daofu,Qimei and Wari) position and CR differential phase in the images.Finally,the observation model for CR points is established to calculate the deformation value of CR points.The leveling results are compared.And the result show that the CR result is consistent with leveling result.
Artificial corner reflector(CR) point is so strong scattering and has stably reflected phase in SAR images that it helps to overcome the limiting factors of coherence in the application of traditional InSAR.At present,the CR-InSAR based on CR points has been used to study the deformation of some areas with complex terrain.We introduce the experiment for CR's installation and identification in the Xianshuihe and Longmenshan fault.Based on the GPS measurement assisted to identify the CR points,a new contrasted method is proposed,which uses the magnitude of the average multiple images before and after the CR installed to identify the CR points.This method can determinate the accurate position of the CR point in the image.
We use disturbance abnormity with features of precursor wave from pointing-precursor observation of epicenter area and nearby within hundreds kilometers in Qinghai-Sichuan area before the Wenchuan Ms 8.0 earthquake on May 12,2008 to search possible epicenter-source area.According to its abnormity vector,extent,time,space-location and the distribution characteristics of tectonic fracture,aided by construction of source-spreading function and inversion of quasi-Newton least square method.Computed epicenter-source is located on the northern segment of Longmenshan fracture,about 134 km away from real epicenter,occurred nearly 13 hours before this earthquake.
We propose a support tensor machine for remote sensing image classification.The training samples are represented as 3-order tensors with local neighbor information.Then the mathematical model and solution of support tensor machine are discussed in detail.A range of experiments demonstrate that the effectiveness of the proposed method can deliver a high classification rate with a small number of training samples.
In dealing with the problem that most current filtering algorithms excessively erode the bare earth at discontinuities or place with complex buildings,we present a new filtering algorithm based on segmentation.Firstly the original point cloud is segmented into many segments based on the continuity of terrain surface.Then,the ground point set is differed from non-ground point set using the way of establishing buffer area after gross error points remove by the number of segment.Finally,we get the filtering effect after several iterations.The experimental results show that this algorithm significantly improves the classification accuracy of ground points compared with other classical filters.This algorithm effectively preserves the ground information when filter kinds of object information.In addition comparing with the filtering effect of Terrascan software,the filtering effect of the proposed algorithm is better in removing vegetation and preserving the point in the edge of ground.
In order to denoise in spatial and spectral dimensions at the same time,a denoising method based on energy variations is provided for hyperspectral image.With the hypersurface minimal scheme,energy functional is built of spatial surface area,spectral curve length and residual l2 norm.The minimal surface area term smoothes data in spatial dimensions and the minimal curve length term works in spectral dimension whose weights are determined by two positive coefficients.The surface mean curvature and curve curvature drive suppressing of noise together.Then perturbation is abated by available neighboring reliable data.The proposed method works well in experiments on hyperspectral image from Hamamatsu camera and AVIRIS.
To the problems about building tilt and ghost image on traditional orthoimage,a novel method for true orthoimage generation based on occlusion detection with TIN(TOGODT) is proposed.The triangle is the dectection unit in this method.Occlusions are dectected by visual sorting of triangles,and occlusion areas are repaired by visible parts of the neighbor images which are treated as weighted image.The experimental results show that this method can accurately detect image occlusion areas,and the generated true orthoimage is complete and has no occlusion.
A registration algorithm which hybridizes the block grids and Delaunay-triangle is proposed to enhance the domestic airborne POLINSAR data accuracy.Firstly,the highest and sec-highest coherence values are used to refine the reliable grid points based on the image blocks.Secondly,the registration error is limited by Delaunay-triangles.Then,the offset model coefficient is solved.The final experiment shows that the proposed scheme increases the POLINSAR images registration.
After analyzing the existing D-InSAR-based DEM construction algorithms,we propose a new DEM extraction algorithm based on phase compensatio.Firstly the simulated topographic phase is remored from the flattened interferogram to reduce its fringe rates.Then,the residual unwrapping phase is compensated with the simulated topographic phase to get the unwrapped interferogram,and refine the baseline estimation and extract height information.Experimental results show that the method can acquire a more accurate DEM over mountainous area.Compared with the conventional interferometric algorithm,the proposed methd can not only improve the accuracy of unwrapped phase,but also obtain a more accurate and significantly improved coherence map to reduce the masked area due to coherence threshold in phase unwrapping.While compared with the height compensation algorithm,the new algorithm can more effectively reduce the height errors due to residual systematic phase errors and stochastic errors caused by residual unwrapping phase and external DEM.
Using the thermal infrared image acquired by thermal airborne hyperspectral imager(TASI),temperature and emissivity are separated.And target fine recognition based on temperature image using multi-scale fractal model was studied.Experimental results show that there is a positive correlation between fractal dimension and complexity of temperature spatial heterogeneity.Spatial distribution of temperature has a character of multi-scale fractal.Multi-scale fractal method was used to extract the thresholds,and extract the edge feature of building successfully.This method has an important value on classification,information extraction,and scale effect.
Crowd counting in dense crowd scene has become an important and difficult topic in the field of automatic surveillance system.Since there are some requirements in the orbital traffic,a crowd counting approach for underground railway station is proposed.A weighted area computing approach is presented according to the information of camera firstly.Then the crowd density is estimated using AdaBoost classifier with the difference of gradient orientation between head and body of people.The people number is counted according to the crowd density and the weighted area.The experiment results show that the proposed approach is effective and feasible for crowd counting in underground railway station.
Measuring distance from images is a common task in engineering task and our daily lives.Distance measurement based on single image which only employs the clues in the image has become a new popular technique.Since circles are common in daily life and process good geometric properties,they has been one of the most important clues for single image measurement.We propose two methods of measuring distance using circles in one single image.Both of them need to compute the projected center of circle and vanishing line of the plane firstly.Then we compute the diameter with the first method,which is parallel to the unknown line segment.After that we can obtain the distance between any two points on the plane based on cross ratio.Under the influence of the location and direction of the line segment to be measured,the errors of the results are obvious diverse in different directions.Instead of diameter the second one compute the exterior of circle.Then the core process of the algorithm is transformed to the distance measurement based on square.Tests on synthetic and real images show that the quality of the second one is better.
According to the time-varying properties of the disturbances which exist in high frequency ground wave radar,we propose a new algorithm of two-dimensions constant false arm rate based on Doppler and range spectrum.The nostationary background is whited with time-varying filter.The three results are compared such as time-varying whitening and two dimension processing,fixed whitening and two dimensions processing,no whitening and one dimension processing.The new algorithm is valued correctly with the comparing results.
Pattern recognition has been a hot issue in the field of map generalization.Grid pattern can be regarded as consisting of cluster of meshes,which have similar properties in location,size,shape and relationship.We present an approach to the recognition of grid pattern in street network using self-organizing maps.Firstly,eight parameters,i.e.centroid,area,rectangularity,elongation,having parallel side,side number,number of 1st order neighbor,mean rectangularity are selected to describe meshes.Each mesh constitutes a vector in attribute space.These vectors then are used to train a SOM.The neurons of a SOM correspond to a set of meshes with similar properties.U-matrix is used to visualizing the SOM.Grid pattern recognition is based on the clusters identified by the SOM.Two experiments are conducted.The results show the proposed approach is valid in mining the grid pattern in irregular street networks.The limitation and future work are discussed.
The resolution of digital elevation models is a basic parameter to generate DEM.However,those studies about how to decide the resolution of DEM didn't take the points number into account.We analyze RMSE of Elevation of ANUDEM and TIN-DEM,which are generated by the data extracted from aero photos by digital photogrammetric method.The results show that the point density for building ANUDEM with 2 m RMSE of elevation and 5 m grid size should be more than 7 000 points/km2,while for TINDEM,it should be more than 8 500 points/km2 at 1∶100 00 scale level.Moreover,RMSE of elevation keeps stable when the grid size is finer than 20 m and the point density is in the range between 7 400 and 9 400 points/km2.
Web processing service(WPS) published by OGC is an important part of GIS Web Services.Existed WPS standard has an obvious shortcoming that data transfer amount of WPS is large,which limits the application of WPS.Using code transferring to replace the data transferring in WPS,we put forward a new WPS model based on code transferring.This model can avoid large work of data transfer in WPS,and will improve the efficiency of WPS.At last,the experimental results show that the new model is reasonable.
Diamond-shaped pattern search algorithm is a fast motion estimation algorithm in AVS.Through an analysis of it,an optimized diamond-shaped pattern search algorithm is proposed,which exclude the work from the surface to the very detail.Firstly,we initial the matching work from the seven points(which including the center point of search window).Then,we search for other matching points according to some certain principles.The experiment results show that the proposed algorithm can reduce the search number and improve the search spirits while keeping better search accuracy.