2012 Vol. 37, No. 1
From opportunity and challenge of geo-informatization with the internet of things and other technology,the new geographic information era and its five characteristic are analyzed.The traditional surveying and mapping product can't meet current society demand,geo-informatization is the next stage in new geographic information era.The architecture of geo-informatization system,increasing service content and form are researched and designed,and current related services which can be promoted are realized.
Considering the properties of watershed-transformation and support vector machine,a method for classifying polarimetric SAR image is proposed in this paper.First,polarimetric SAR image is segmented into a series of homogenous regions through watershed transformation and region merging process.Then,region-based classification is performed by utilizing support vector machine after feature extraction and sample selection.Experimental results show that the proposed classification method depresses speckle effectively,when in comparison with traditional pixel-based SVM algorithm,the classification accuracy is improved by dramatically and more interpretable result can also be achieved.
In order to solve the problems that conventional SIFT descriptors result in a lot mismatches when an image has many similar regions,an improved SIFT Descriptor and its generation method are proposed.Firstly,the steps of constructing scale space,detecting extrema and confirming keypoints are employed to create the key feature points.Secondly,each keypoint is described by the rotation-invariant Texture patterns,which are computed from the image patch centered at the keypoint,and be combined to SIFT feature to construct the union feature.Finally,a rotation invariant textured SIFT descriptor is formed.Two different matching strategies are adopted as the method applied to image matching.The experiment show that the method makes the image content information more comprehensive,it enhances the accuracy of matching and improves matching results greatly,achieve the improvement of SIFT algorithm.
This paper presents our improvement to the Tsai's monoview calibration approach with enhanced ability to principle point issues and higher calibration precision is achieved as well.Taking one grid plane as reference object,well-known radial alignment constrain(RAC)is extended to handle with principle point position parameters firstly,though function-dependency between principle point and other parameters are introduced.Next,vanishing points intersected with the perspective projection of grid lines,parallel to world coordinate system X,Y axis respectively,are related to dependent parameters involved in calibration equation from extended RAC and as a result,new non-linear calibration equation with independent parameters is deprived.After that,procedure using improved approach to calibrate camera parameters is specifically listed.In this step,one Least Square based iterative solution is proposed to solve the non-linear calibration equation as well as initial parameter values are given.Finally,stimulated images generated with OpenGL toolkit with known camera parameters and real image from CCD camera Nikon P5100 are used for calibration experiments.Valuable conclusions are conducted.
It's suitable to distinguish the scattering properties of different objects basing on Wishart-H/Alpha classification.However,various scattering mechanisms are merged when class clustering,which confuse the road and the bare soil.In this paper,an improved scheme is performed which segments the optimal-coherence and Polarimetric-Span 2D-histogram to subdivide the road and bare soil after Wishart-H/Alpha classification.Finally,the experiment on domestic dual-antenna X band POLINSAR datum proves that proposed method can separate roads and bare soil effectively.
Traditional multi-look SAR images are highly sensitive to target aspect angle,which is unfavorable to target recognition.To solve the problem,a static modeling method for multi-look SAR images is proposed.The method integrates images from multiple aspects into a composite data structure.The data structure is reorganized so as to be function of target scatter centers,not of aspects.Then the static model is used to modeling the target data of incomplete aspects,and form the template of different targets.The input multi-look SAR images are classified using template matching algorithm.The theory analysis and simulation result show that the method excels traditional model when a small quantity of SAR images of different aspects are feasible.
As image surface is introduced for the surface Min/max flow denoising method,the curvature of isolux curve in classical Min/max flow is replaced by the one of image surface and improves switching mechanism for better adaptability on complex noise,while adding gradient weight to control evolving speed of image surface for better edge preserving.The proposed method makes Center-Weighted Mean filtering as preprocessing in order to eliminate remarkable impulse noise which could make inefficiency in Min/max flow.In the denoising experiment of Remote Sensing image with artificial noise,the resultant image indicate that the proposed method removes noise effectively while preserving edge well,and the statistical data for PSNR,SSIM and detail preserving index is good or better than some other popular denoising methods such as BM3D,wavelet.
In order to avoid disadvantages of the current algorithms,the Particle Swarm Optimization(PSO) algorithm in the field of artificial intelligence is successfully applied to remote sensing image segmentation.And a new algorithm combined PSO and Isodata is proposed in this paper.This method first changes the color space of the images,then the initial cluster number are determined by the combined algorithm.Finally,the automatic segmentation of remote sensing images is achieved through multiple iterations.Through many experiments of remote sensing images with different spatial resolution,the results show that the new algorithm can determine the initial cluster number adaptively,avoid the local optima of K-means and Isodata algorithms,increase the searching capability of PSO,and the segmentation results are much more close to the actual situation.So it is a new effective algorithm for the segmentation of remote sensing images.
In this paper,a new method of quasi-automatic extraction of zonal roads is put forward.Based on research on roads' feature in high resolution RS images,this paper uses two seed points chosen on road surface to automatically search original information of roads;Then,we can establish standard road template,and match it with new road templates got by moving and rotating standard template to compute new direction,center point and edge points of next road segment;Through constantly updating road standard template,direction and center point,automatically obtain a series of edge points distributing on the two-edge of road;Use Snake Model to dynamically adjust edge points and accurately position them on the true edge;Lastly,solve B-spline curve of control points to denote the two borders.The results show the research is effective and viable;Meanwhile,the research makes great advance in the degree of automation,efficiency and accuracy.
A registration method for SAR image based on cascade filter and relaxation optimization algorithm is proposed in this paper.Firstly,the cascade filter method combining partial differential equation(PDE) and enhanced Lee filter is introduced for SAR image speckle reduction.Then edge extraction is done in the filter result of object image.Every feature point of the edges is matched by the relaxation optimization algorithm,a pyramid matching strategy is employed for the accuracy simultaneously.The experiments show that the method can reach a good registration result.
In this paper,we propose a novel hierarchical classification algorithm based on feature selection and adaptive decision tree in SAR image classification.Firstly,Joint Boosting selects feature combination most suitable for each class;Secondly,a hierarchical classifier is searched adaptively using binary classifiers based on feature combination;finally,we perform SAR image study and inference based on feature selection results and adaptive decision tree,leading to automatic classification.Experimental results on the first batch of PolInSAR data prove the proposed approach's efficiency.
In order to improve the performance of morphological edge detection algorithm,a compound filter with multi-structure elements was designed.At the same time,a morphological gradient with orientation was defined.On the basis of that,a noise immune morphological edge detection algorithm was presented.Noise suppression and detail preservation capability of morphological transformation mode and structure element were developed sufficiently.Edge was detected with structure elements which possessed direction information,and it was thinned with one pixel width by implementing non-maxima suppression along the orientation of gradient.Simulation results indicate that the proposed algorithm not only performs better in edge detection,noise immunity and processing speed.
According to the different sampling rate or different scale of different navigation sensor,this paper puts forward one information fusion algorithm for asynchronous multi-sensor integrated navigation system based on the multi-scale transformation of state equ-ation.This paper first builds up the state equation of integrated navigation system based on the highest sampling rate or the finest scale,then this state equation is decomposed into different scale to establish several state equations based on different scale and the corresponding measurement equation,at last the global and optimal information fusion algorithm based different scale is finished.The simulation results show that this algorithm has not only better real-time,but also better fused precision.
Here we introduced a new kind of ionospheric model(SEID),and it is adapted for near real-time weather monitoring based on the characteristics of data processing for large network.The effects of this strategy in different size of reference network have been studied through serial analysis of a example of SXCORS.For reference network with averaged separation shorter than 249 km in Shanxi Province,we can gain regional PWV within 1~2 mm using single frequency observations.
Accuracy analysis is the precondition of a system's production accuracy ensure,through which it is helpful to determine the systemic performance indexs and some technical parameters in order to make the system design easily.In the article,combined with the SAR imaging principles,especially on the basis of presenting the solution of DEM's three dimension coordinates in airborne InSAR system with DGPS/IMU data,deduced the DEM precision influence formulas caused by several errors factors such as base line length,base line inclination,phase difference,distance from the SAR antenna to the ground target points,SAR antenna's position,center Doppler and the plane's velocity,in order to improve the systemic precision then analyzed the main influencing accuracy factors in airborne InSAR system and the systemic realizable performance condition,meanwhile proposed some design suggestions of the airborne InSAR system and made experiments on the real airborne InSAR data.In the article,the results indicated from the methods of accuracy analysis is accordant with the statistical precision using real experimental data,and the conclusions influencing accuracy analysis are correct.
Inertial navigation system fine alignment based on Kalman filter is normally carried out under navigation frame.Earth-centered-earth-fixed(ECEF) frame is an important frame for combination navigation and inertial geodesy but no previous research has been shown for the fine alignment under this frame.Beginsing from mathematical deduction of a 12-states Kalman filter equation under ECEF frame,we estimate the misalignment angles and determine the transformation matrix from body frame to ECEF frame for the fine alignment.A simulation calculation is accomplished for this algorithm to prove its validity and availability.Besides,the alignment accuracy is also discussed.Experimental results show that the fine alignment can also be done under ECEF frame for inertial navigation system.
In order to avoid the conventional interpolation model unrelated with observations and improve the accuracy of network RTK atmosphere error interpolation,a new model suitable for atmosphere error estimation with large height difference is given based on atmosphere mapping function and conventional interpolation model,whose variables are satellite elevation,reference station height and distance respectively.Finally,different interpolation models are used to process observations of part reference stations of Jiangsu CORS for estimating atmosphere error.The results by comparison and analysis show that the new model will increase the accuracy of atmosphere error estimation effectively.One hand,mapping function estimation method and lower-order surface model can accurately estimate the double difference troposphere error between users and main reference station with less than 2cm in maximum mean square error.On the other hand,mapping function estimation method,lower-order surface model and linear interpolation method are equivalent in accuracy for double difference ionosphere estimation,but the accuracy of distance interpolation method is the worst with nearly 5 cm error.
The remove-restore process essentially is to replace the data outside the observed area on Earth with some known model.But in practical application,there are some problems needed to be studied when the remove-restore process is applied.For example,how to determine the integral domain and to evaluate the accuracy Firstly,this paper gives a new derivation of inverse stokes formula,and then concretely analyze and discuss the accuracy of the remove-restore method in order to research the source of error in actual computation.Combined with the accuracy problem,we focus on analyzing size of the integral domain and the number of the removed degree and obtain several useful conclusions.
Owing to the fine resolution of TerraSAR-X data provided since 2007,this paper applied 6 TerraSAR data(strip mode) during 3rd Dec.2009 to 23rd Mar.2010 to detect and monitor the active fissures over Xi'an region.Three themes have been designed for high precision detection and monitoring of Xi'an-Chang'an fissures,as small baseline subsets(SBAS) to test the atmospheric effects of differential interferograms pair stepwise,2-pass differential interferogram with very short baseline perpendicular to generate the whole deformation map with 44 days interval,and finally,corner reflector(CR) technique was used to closely monitor the relative deformation time series between two CRs settled crossing two ground fissures.Results showed that TerraSAR data are a good choice for small-scale ground fissures detection and monitoring,while special considerations should be taken for their great temporal and baseline decorrelation.Secondly,ground fissures in Xi'an were mostly detected at the joint section of stable and deformable regions.Lastly,CR-InSAR had potential ability to monitor relative deformation crossing fissures with millimeter precision.
To take advantage of geometric and physical information of slopes and restrain the influence of the observation outliers on the estimates of deformation parameters,a filtering model handling unknown systematic errors is developed.An adaptive fitting algorithm for the systematic errors based on moving windows is presented and the estimation method for covariance matrices of the predicted states is given.The presented algorithm utilizes the statistical information of observations as well as landslide related information such as mechanics status and geological conditions.The results of the GPS monitoring network show that the algorithm may reduce the effect of abnormal observation by fitting the geophysical information,and therefore improve the precision of deformation parameter estimates.
Normal detection mode using large area grid data is not applicable to mobile underwater target's localization and classification with airborne technique.Therefore,a new detection mode was proposed here only require magnetic profile lines data around the body,and the its location and physical parameters were calculated with an improved least square(LS).For assuring the astringency of calculation result,the parameters are divided into linear parts and non-linear parts.In the calculation process,two parameter parts were figured out respectively with only three unknown parameters in non-linear equations.Considering the influence of background field and multi-sources situation,corresponding data processing methods were also put forward.The effectiveness of the suggested techniques has been illustrated by real magnetic data from a collection of environmental ferro-metallic objects.The conclusion shows that the presented approach is convergent and the calculated geometry together with physical parameters has very high precision.
Spatial clustering has been a major research field in spatial data mining;it aims to discover some useful patterns or outliers in a spatial database.In practice,spatial obstacles,as river or mountains should be fully considered in the process of spatial clustering.On that account,a novel spatial clustering method considering spatial obstacles is proposed in this paper.Delaunay triangulation is employed to model spatial proximate relations among entities,and the method can automatically discover clusters with complex structures without user-specified parameters.Experiments on both simulated database and real-world database are utilized to demonstrate the effectiveness and advantage of our method.
This paper improves Isomap algorithm,which is a nonlinear dimensionality reduction algorithm,and proposes a spatial alternation method based on metric multidimensional scaling.This method transforms road network distance in the original network space into approximate road network distance in a new Euclidean space,and then achieves Kriging based on this distance metric.The experiment of Nanchang's real data shows: this method has higher estimation accurate than Kriging based on Euclidean distance metric.Therefore,it is an effective solution to the problem of large-scale-road-network-level traffic state monitor.
In the paper,taking Xiannangou watershed,representing the loess hilly geomorphology landscape,as the test area,a series of hydrologically corrected DEMs with grid cell from 5 m to 200 m were generated using ANUDEM based on the 1∶10 000 digital topographic map.Different resolution slope gradients were extracted within the Arc/Info slope calculation methods.By applying terrain analysis and geostatistical methods,changes of DEM-derived slope with horizontal resolution and their spatial distribution were systematically investigated,both on point,line and surface level.Research shows that with DEM resolution decreasing,slope at single point shows uncertainty,while points at the same slope level change regularly-low slope terrain increases at first and then decreases,middle slope terrain shows slight changes,and steep slope terrain decreases with logarithm greatly.
An automatic method for viaduct detection with LiDAR data and remote sensing image is presented in the paper,according to spatial characters of urban complex viaduct.The flowchart and key technologies of the method are discussed in detail.The feasibility and performance of the suggested techniques are verified through experimental results with real data.
To maintain shape similarity of features is one of evaluation criteria before or after in map generalization.The article derives shape similarity of line-feature from shape descriptor of polygon——rebuilds conjugate polygon of line-feature consisting of mirroring-part of original feature using the axis of symmetry through the first point and the end point of itself,gets a shape vector of the line feature,which is the shape descriptor of its conjugate polygon and further requires the similarity distance among line-features through Euclidean distance between two vectors.To ensure certain evaluation accuracy,the K value is determined by calculating Kth item X,Y value of Fourier series.In the end,the shape fidelity of several simplification methods to given curve is analyzed to provide the reference to select which algorithm will be used.The model is simple,with mathematical rigor,and highly practical.
This paper advances a concept of structure of ecosystem service value,introduces geo-information tupu and takes Jiangxi Province as the study area.Firstly,the various types of land use and their areas for each county in Jiangxi acquired by classifying the HJ remote sensing images.Then,the ecosystem service values for six type of land use are calculated by the modified estimation method of ecosystem services value.Thirdly,the ecosystem service values for counties are classified to different types based on the weaver combination index.Finally,combined with ecosystem service value types and geo-information tupu,and ecosystem services values genealogy of Jiangxi province are acquired.The characters of the structures and their combination types of ecosystem service value in Jiangxi province are analyzed following the above procedure,in order to give scientific suggestions for sustainable development of regional ecological environment,Low-carbon Jiangxi establishment,and ecological economy research.
This paper analyzed the common electronic map projection distortion,and proposed a discontinuous change of the standard parallel for map tiles' indexing method based on the equidistant cylindrical projection.This indexing method sets angle deformation at the standard parallel as 0 and reduce the deformation by custom defined standard parallel and changing it automatically.Through computing and comparing the length deformation,angle distortion and area deformation of this method and others,it shows that this method have moderate deformation characteristics and also have good deformation controllability in regional electronic map.Therefore,this indexing method will have a high practical value.