2013 Vol. 38, No. 12
The efficient feature detection is a crucial step for various tasks in computer vision. However, the corner location is not very accurate and some edge pixels have high responses for the kind of corner detector based on the accelerated segment test. Furthermore, this kind of method is sensitive to the change of illumination. This paper presents a robust and efficient feature extraction transform (REFET). The method includes three steps: firstly, the input image is enhanced based on the color constancy of the human vision system. Then, acoarse edge pixel detection method is used to eliminate edge points in the image. Finally, an adaptive mask is proposed to extract corners accurately and efficiently. The experiments demonstrate that the proposed method improves the corner localization accuracy and the robustness for illumination, and the edge points are suppressed in the results.
It is been an important topic in the field of astrometry and geodesy to monitor thereference point and axis offsets of VLBI antenna with high precision.The common monito-ring method is based on some special restrictions to the rotation mode of antenna,which ab-sorbs much of the effective time during operation of the telescope.Monitoring is also subjectto meteorological conditions,and so there are limitations in the actual precision and timeli-ness of such monitoring.In recent years,the concept of VLBI antenna monitoring based onparameterized modeling of the antenna rotation has gradually become more acceptable,and isexpected to perform all-weather and fully automatic monitoring of the antenna without anyinterruption of normal operation of the telescope.In order to investigate the feasibility of thisconcept when designing monitoring procedures and to develop data processing software,some insights and analysis concerning the observation equation,the settings of unknown pa-rameters and the application of constraint conditions are presented.A simulation analysis ofthe VLBI antenna monitoring using GPS data and some preliminary conclusions are given forcriticism and reference for readers.
There exist two main categories of attitude determination approach using globalpositioning systems.The first is based on a single baseline observation model,which choosesbaseline and ambiguity as its parameters.Firstly,the baseline coordinates were estimated inagiven frame.Then,in alliance with the corresponding baseline coordinates in the rigid bodyframe,the attitude was estimated.The second category is based on mutibaseline observationmodel,which takes ambiguity and attitude matrix as its parameters,and solves the ambigui-ty and attitude matrix simultaneously.The CLAMBDA method and MC-LAMBDA methodwere choosen as the representative algorithms of these two categories of approaches,andthey are described in detail,then a comparation and analysis between them were presentedwith numerical examples.
We first introduce GNSS precise point positioning(PPP)based on un-differentialuncombined raw pseudo-ranges and carrier phase observations(RAW-OBS-PPP).Then thedifference between the traditional PPP based on ionosphere-free dual-frequency combination(LC-PPP)and RAW-OBS-PPP is discussed and evaluated by analyzing mass data of 300IGSGPS sites during day of year 034～０４０,2012.The effects of receiver’s DCB and the preci-sion of priori-ionospheric delay on the convergence of RAW-OBS-PPP are focused and dis-cussed in details.For RAW-OBS-PPP,while global ionosphere mapping(GIM)provided byInternational GNSS Service(IGS)are used as priori-information for ionosphere delays,andthe receiver’s DCB is estimated,then the mean time is about 13.12min to take convergenceinto 10cm in both north and east components.If no receiver DCB estimated,it needs about18.48min.LC PPP and RAW-OBS-PPP have almost the same performance under the condi-tion that GIM is used for RAW-OBS-PPP.If ionosphere delays with higher precision areused as priori-information,RAW-OBS-PPP needs about 8.67min(mean value)in horizontalcomponents,which is much faster than traditional LC PPP.
Ocean tide loading has a crucial geophysical effect on GPS precise positioning.Based on the theory of ocean tide loading,ocean tide loading displacements for IGS stationsin China and surrounding areas resulting from four global ocean tide models are calculated,and their influences on GPS daily and sub-daily solution are analyzed and compared.For dai-ly solution for the station,results show that ocean loading displacement corrections lead tothe same RMS for the daily solution residuals with no corrections on average,but for RMSsof daily solution residuals in the vertical direction,there are changes in coastal areas,so it isbelieved that four ocean tide models have insufficient accuracy for the offshore area of theYellow Sea and the East China Sea,and must be corrected by the data for the China offshoreregional ocean tide.For sub-daily solutions for the stations,results show that the ocean tideloading effect decreases with the distance of the station from the sea.Particularly in coastalareas,the accuracy of the vertical coordinates of the station can be significantly improved,soocean tide loading displacement corrections should be considered in sub-daily GPS precise po-sitioning.
As an important part in the near-Earth space environment,the ionosphere activityis closely related to human survival and development and has a significant impact for produc-tion and life of human beings.At present,most studies are based on the ionosphere iono-spheric total electron content(TEC).It can only reflect the level of the ionosphere struc-ture,but cannot reflect the vertical structure of the ionosphere.Computerized ionospherictomography(CIT)technology based on GNSS is most convenient and effective.AnalyzingCIT techniques,some conditions are proposed to prove its accuracy and reliability.Using ob-servation data provided by IGS stations on August 3,2010,an ionospheric image of the Eu-ropean region is reconstructed.It shows the feasibility of this method.
Compass Geostationary Earth Orbit(GEO)satellites observations are used tostudy medium-scale traveling ionospheric disturbances(MSTIDs).Themethod to exactionos-(下转第1429页)1412 第38卷第12期 翟振和等:利用级数展开法确定区域大地水准面的重力位Local Geoid Gravitational Potential DeterminationUsing the Series Expanding MethodZHAI Zhenhe1,2 SUN Zhongmiao2 WANG Xingtao2 LIU Xiaogang2(1 Institute of Geography Spatial Information,Information Engineering University,66Middle Longhai Road,Zhengzhou 450052,China)(2 State Key Laboratory of Geo-information Engineering,1Middle Yanta Road,Xi’an 710054,China)(3 State Key Laboratory of Astronautic Dynasmics,28P.O.Box 505,Xi’an 710043,China)Gravitational potential can be expressed as a mathematical series,in which the firstsection is geoid potential-W0.This method was used to determine the Baltic Sea geoid poten-tial by using the GPS station data(1997-04).The results show that the first rank term com-puted by EGM2008is almost equal to the results computed by WGD2000(complete to degreeand order 360)while the difference is about a quantity of 10-3.The second rank term has aquantity of 10-5and can be omitted.In the case of changeless geocentric coordinates,the de-termination of W0is largely affected by changing height data,so it is important to accuratelydetermine the height of the GPS station.The W0value(62 636 857.23±０．４８ｍ2·s-2)ofBaltic sea study site is obtained by using five set of GPS stations and an EGM2008modelwith degree and order of 360.
New composite Global Navigation Satellite System(GNSS)signals are usuallycharacterized by the presence of data and pilot channels and the adoption of a tiered codestructure.The problem of acquiring composite GNSS signals of the next generation systems(such as European Galileo,and U.S.s GPS modernization)is addressed.Several joint data/pilot channel combination strategies have been proposed to effectively deal with data and pilotchannel combinations in order to make full use of the transmitted signal power available inboth channels.These channel combination strategies were further updated with a differentialdetection scheme to improve the robustness of the acquisition process of GNSS receivers,particularly in weak signal environments.The statistical properties of differential channelscombination strategies are presented in detail.The performance of the proposed techniqueswas evaluated using Monte Carlo simulations on simulated Galileo E1Open Service(OS)sig-nals.The results reveal that differential channel combination techniques provide improvedperformance in comparison to the state-of-the-art single channel acquisition methodology.
The cosine trend model and ARMA model for pole motion have been fitted andforecasted under the use of the data offered by IERS.Through the analysis of the experimen-tal results,we find the cosine trend model more suitable for analyzing the components of thepole motion.Meanwhile,ARMA model is more suitable for fitting pole motion time series,its fitting accuracy is higher than the cosine trend model and its prediction accuracy havereached an advanced level.
Gravitational potential can be expressed as a mathematical series,in which the firstsection is geoid potential-W0.This method was used to determine the Baltic Sea geoid poten-tial by using the GPS station data(1997-04).The results show that the first rank term com-puted by EGM2008is almost equal to the results computed by WGD2000(complete to degreeand order 360)while the difference is about a quantity of 10-3.The second rank term has aquantity of 10-5and can be omitted.In the case of changeless geocentric coordinates,the de-termination of W0is largely affected by changing height data,so it is important to accuratelydetermine the height of the GPS station.The W0value(62 636 857.23±０．４８ｍ2·s-2)ofBaltic sea study site is obtained by using five set of GPS stations and an EGM2008modelwith degree and order of 360.
Based on the least-squares method,the area of Lunar Sinus’gravity field is recov-ered using the line-of-sight acceleration data from Lunar Prospector extended mission.Bycounting the least-square residuals and calculating the power spectrum,as well as comparingthe gravity model of line-of-sight with the high resolution gravity models,such as SGM150j,GLGM3,LP150Q,the feasibility and reliability of the line-of-sight method in improving lu-nar local gravity models is verified.The correlation between topography and gravity of Sinusis also briefly analyzed.
We propose a generalized inverse distance weighting method after discussing theproperties of the Taylor series expansion of the traditional inverse distance weighting func-tion.Our generalized inverse distance weighting method is established by a set of virtual ob-servation equations from the Taylor Series expansion of the spatial function.The probabilitymeasure,defined by the variance-covariance matrix,and the k-order partial derivatives esti-mation,is used to determine the weights for virtual observations.In order to optimally de-termine the parameter dimensions for the model,the criteria of BIC is introduced.The appli-cable conditions for the traditional inverse distance weighting average method are obtainedfrom the first-order generalized inverse distance weighting average method.At last,the pro-posed generalized method is applied to a GPS leveling fitting problem to verify the proposedmethod.
Proper equality constraints between parameters can improve the accuracy of a leastsquare solution,eliminate the deficiencies of free networks and improve the solution of an ill-posed problem,therefore we proposes the ill-posed adjustment model with equality con-straints.First,we established a regularization rule and developed a solution through the La-grange multiplier method.Next,we evaluated its statistical properties to show that this so-lution is biased and has a smaller mean square error(MSE)than that of the constrained leastsquare problem solution.Then we propese a rule that minimizes the MSE of the estimates tochoose both a regularization factor and numerical algorithm.Finally,we use a simulated ex-ample and another control net to verify the feasibility of the algorithms and test the statisticalproperties of the solution.
For boundary pollution problems occurring during the process of SAR image noisesuppression by bidimensional empirical mode decomposition(BEMD),an improved BEMDmethod of image denoising is proposed.In this method,the noisy SAR image is decomposedinto detail parts and contour parts using BEMD.First,the extreme points are triangulated,and interpolation boundaries equivalents are extended.Then,the surfaces of the IMF com-ponents after interpolation are decomposed,and the IMF components with noise after the de-composition are processed by wavelet filtering.Finally,images are reconstructed.Experi-mental results show that this method effecive for suppressing speckle noise,as compared tothe bidimensional empirical mode decomposition,and has sufficiently advantages when retai-ning edges and detail information while suppressing noise.
A rigorous geometric processing model for ZY-3satellite images based on image-forming principles for three-line array sensors equipped in ZY-3was built using satellite statedata such as the line-of-sight angle.The validation results show that positional accuracy wasnot good when satellite state data and a rigorous geometric processing model was directly ap-plied without processing.This was caused by errors created when obtaining satellite statedata such as the line-of-sight angle and sensor position.In order to decrease original systemerrors in satellite orbital position and attitude angles,a modified model based on the line-of-sight vector was introduced to modify these system errors.After the processing,positonalaccuracy in flat areas was more precise than 3meters even when only 3control points wereincluded in x-y directions.
In the processing of mass remote sensing products production,the influence resul-ted from different radiometric and geometric correction orders has not been analyzed overall.So it is highly meaningful to quantitative analyze of the images discrepancy caused by differ-ent radiometric and geometric correction orders.It will help to clear the influences caused bythe changing correction orders,and also play a significant role on improving the accuracy ofquantitative products.We used HJ-1Bsatellite CCD data as fundamental datum and pro-cessed them.Firstly,integrating data characteristics,we determined methods of radiometricand geometric correction.Secondly,technology roadmap was designed by changing the radi-ation and geometric correction order.Finally,the results obtained by different orders of ra-diometric calibration and geometric correction were quantitative analyzed from two aspects ofstatistical characteristics and texture features of images.The experimental results showedthat in the statistical and texture indexes of results,most of them had little difference.Theexperimental results can prove a fact that different correction orders have a little influence onthe pretreatment results.
As one of main classification methods used in data mining,the decision tree algo-rithm is widely used in remote sensing image classification.However,in current studies ofremote sensing image classification,the building of decision trees was found to be dependenton existing data mining software,with little research work focused on decision tree algo-rithms.Based on the BoostTree algorithm,we propose a new algorithm of decision tree en-sembles for remote sensing image classification-AdaTree which is a combination of C4.5andAdaBoost.M1algorithms.In AdaTree,the structure of C4.5and the final hypothesis of Ad-aBoost.M1were modified.With the AdaTree classifier algorithm,apiece of software wasdeveloped for cell-based and object-oriented remote sensing image classification.An experi-ment with Landsat7ETM+ and Wordview2images showed accuracy and efficient improve-ments of the AdaTree classifier when compared with BoostTree and SVM,either in cells-based or object-oriented classification.Its average Kappa coefficients reached 0.905 2and0.939 8.
Relying on the Parseval energy conservation theorem and‘Theory of Tupu’，thefeasibility of object recognition methods based on‘pixel-baesd spectral energy’to‘feature-based spectrum energy’was demonstrated.A preliminary investigation of the percentage ofspectral energy contained by circles of different radius taking total energy,and the spectralenergy distribution was undertaken.Meanwhile,aquantitative analysis of coarse-grainedlevel,cyclical,and directional information was studied using wedge-shaped and ring-shapedenergy spectrum sampling methods.Finally,based on low-frequency and high-frequency rec-ognition marks,residential buildings were taken as an a test-case to assess this theory andmethod.Building objects were extracted with matched Gabor filters having direction and fre-quency selectivity.The results indicate that this method provides a new and effective ideameans further identify and extract information from high resolution imagery.
As compared to a pixels’regular neighborhood system,the morphology and distri-bution of segments are irregular,which makes the quantity of neighborhood cliques in RAG-MRF unpredictable,and therefore the quantification of segments’orientation features be-comes impossible.Aimed at the disadvantages of current segments-based random field in ori-entation feature,we propose a random field model based on segments spatial integrated adja-cent potential energy analysis,a new neighborhood potential energy function based on statis-tic was used,and according spatial adjacent feature indices for segments was designed,andthe results of classify or recognition was improved.
The joint fingerprinting and decryption(JFD)framework can provide comprehen-sive security protection for distribution of remote sensing images,and is highly efficient.But,this framework will lead to degradation of image quality after embedding fingerprints,and therefore cannot meet the specific near-lossless requirement of remote sensing images.Asecure distribution method for remote sensing images based on content is proposed.The im-age is partitioned into different regions,and the region which affects image content greatly isexcluded from the fingerprint embedding area.At the same time neighborhood similarity isdefined to evaluate the change of pixel correlation,areas with good neighborhood similarityafter encryption are chosen or embedded fingerprints.Experimental results show that thefingerprinted image has good image quality,and it will not affect its applications.
According to the Earth observation mission requirements,an observation require-ments model based on technical indicators for imaging-satellite sensors was determined.A-dopting the fuzzy comprehensive evaluation(FCE)method,a novel quantitative method of e-valuating imaging-satellite sensors’observation capabilities in regard to mission requirementsis put forward.The method can make reasonable assessment of the applied observation capa-bilities of imaging-satellite sensors.Experimental results show that metric imaging satelliteobservation capabilities using the proposed method are feasible.
Regionalization based on the elements of climate using a clustering method is animportant approach to identify climate zones.A scale-space clustering method is proposed formulti-scale regionalization of sea surface temperatures from the Pacific Ocean.First,a modi-fied clustering method is developed based on scale-space theory for climate time sequences.Second,the method is used to regionalize the sea surface temperature of the Pacific Ocean.The results obtained demonstrate that the proposed method is able to reconstruct the climateindices accurately without input parameters,and is more effective and advantageous thanother approaches.
To solve the quality problem of crowd sourcing geographic data obtained from non-professionals,aquality analysis model for open street map(OSM)crowd sourced geographicdata is proposed.Aimed at the data characteristics of OSM,an OSM spatial data qualitymodel with a calculation method is presented with data completeness,attribute accuracy,andpositioning accuracy as three quality elements.
Massive three-dimensional GIS(3DGIS)data rendering is computationally inten-sive,and time-consuming.Therefore to address these limitations,we used OpenMP paralleltechnology to design a multi-core 3DGIS parallel rendering model based on multi-core com-puting and programming.The purpose of the model is rendering and visualizing 3DGIS databy deriving multiple simultaneous threads to take full advantage of multi-core computing re-sources.Finally,using an instance of a three-dimensional urban scene we verify the calcula-tion of speedup by multi-core parallel rendering.
Landslides are major natural geological disasters in China,and large-scale engi-neering activities induce and aggravate the occurrence of catastrophic landslides.Traditionalspatial analytical techniques cannot easily discover patterns,trends,and relationships thatcan be hidden deep within complicated landslide hazard systems due to limited data sourceand long update cycle.Focusing on the Three Gorges,a variety of environment and trigge-ring factors for landslide occurrence were calculated or extracted from the multi-source spa-tial data.Secondly,the study area was partitioned into slope units derived semi-automatical-ly from a digital elevation model to resample the conditioning factors.Finally,a two-classSVM was trained and then used to map landslide susceptibility with the best accuracy of 98.21%.To evaluate the models,the susceptibility maps were validated by comparing themwith the existing landslide locations according to success rate curve and error rates.
Aiming to improve the small file problem of a general distributed file system incloud storage,a file merging and prefetching strategy based on a users’access task is pro-posed that improves the PLSA model.Analyzing the relationships among the access tasks,applications,and access files the strategy merge small files on the basis of tasks and selectsprefetching files based the transition probability of the tasks.Efficiency model analysis andexperimental results of a digital city prototype system based on HDFS all show that the pro-posed strategy has a high prefetching hit ratio and can effectively reduce the metadata serv-ers’load and the response delay for users’request.