2013 Vol. 38, No. 7
After reviewing the existing true ortho image rectification methods,we analyze image rectification and texture resampling processes that use a pixel-oriented strategy.These strategies generate inaccurate geometric features,imperfect texture structures of man-made objects.Thus,occlusion recovery and shadow compensation are difficult to automatically process,requiring much manual intervention.Aiming to solve these problems,this paper proposes a novel object-oriented true-ortho rectification method,implementing a high-level object-oriented strategy instead of a low-level pixel-oriented strategy.There are three main steps:① definition of physical objects and image object containing geometric and semantic information.In 3-D physical space,all the objects in the horizontal plane of projection form a seamless surface comprised of a series of contiguous triangular facets,in this paper this surface is represented by Semantics constrained Triangulated Irregular Network(STIN).The physical objects are characterized by different aspects: property,outline,geometry and topological relation,extracted from 3-D city models and point clouds.The image objects are made up of several pixels with semantic descriptions for spectral information,2-D geometry and topological relations.They can be extracted by image segmentation,edge detection,and texture clustering.② establishment of a global hierarchical spatial index of image objects.After deriving visual correspondence relationship between 3-D physical objects and 2-D image objects,we use the index to organize all the information from physical and image objects efficiently and provide a foundation for high performance true-ortho rectification and optimized image sampling.③ object-oriented true-ortho rectification and optimal image resampling.True-ortho rectification is carried out based on the physical objects as STIN.And based on the semantic links in the global index,the image objects with semantic information on visibility,integrality,and radiometric features can be chosen for optimized sampling to adaptively handle occlusion and shadows.Experimental results revealed that the proposed method has the advantages of high accuracy,high efficiency and full automation.
Road characteristics are not fully used in road extraction methods based on pixels.In this paper,an object-oriented road extraction method based on road-unit is proposed.First,a road-unit is defined according to the road characteristics.Then,searching,construction,tracking and connection of road-unit were completed with region growing support.The preliminary road network was extracted and a method to repair the road information was designed to obtain the final road network.GeoEye satellite image experimental results validated that the proposed method is practical and effective.
This paper proposes a registration method based on object-space positioning consistency for satellite multi-spectral images.The method includes two steps: restoring the precise band-to-band relative geometric imaging relationship by adopting a method similar to relative orientation,and realizing automatic registration with sub-pixel accuracy without image-matching using the constraint conditions of object-space positioning consistency of the homologous image points.The method uses a rigorous geometric imaging model as the rectification model,achieving true geometric registration and is theoretically strict.The results,furthermore,have no relationship with the image quality.In our experiment,ZY3 satellite multi-spectral images are tested to demonstrate the feasibility and correctness of our method.This method has been successfully applied in the satellite image pre-processing software.
A modified Speeded-up Robust Features(SURF) descriptor based on Haar-like features,line features and diagonal features are added in the descriptor is proposed.Integral images allow fast computation,and are used for matching with SURF.However,the SURF descriptor has a lower discriminating ability when compared with SIFT.Experimental results demonstrate that the modified descriptor has a better discriminating ability compared with SURF descriptor.
In this paper,we propose an improved spatial-weighed fuzzy C-means algorithm since the traditional fuzzy C-means algorithm is more sensitive to the initial cluster centers than other commonly deployed algorithms,such as the FCM,SFCM algorithms.BP neural network algorithms are used to train samples and obtain an initial membership matrix,thus increasing the reliability of the initial cluster centers.Since spatial data have pattens of spatial auto-correlation,the neighboring pixels will contribute to the center pixel with different weights to robustly handle noises.Segmentation experiments were conducted using SPOT 2.5 meters remote sensing images to verify the effectiveness of our algorithm.In comparison with FCM,SFCM algorithms,the experimental results show that the proposed method obtains better results.
An image denoising method based on the non-subsampled Shearlet domain Gaussi- an scale mixture model is presented.First,a Gaussian scale mixture model is used to model the correlation of the locally non-subsampled Shearlet coefficients of the noisy image.Then,the noise-free coefficients are estimated by the Bayes least square estimator.Finally,the inverse non-subsampled shearlet transform(NSST) is applied to these estimated Shearlet coefficients to obtain the denoised image.Experimental results show that the proposed method can remove Gaussian white noise while effectively preserving edges and texture information.At the same time,it can achieve a higher PSNR and mean structural similarity than the wavelet based GSM method,the curvelet domain multivariate shrinkage method and the non-subsampled Shearlet domain hard thresholding method.
While traditional linear and non-linear filters reduce the noise in images,some important details are unavoidably lost,especially linear structures such as curves and edges.Morphological extension is a valid measurement mechanism to distinguish noise and detail information.Different structure features can be expressed by measuring their morphological extension.Non-linear filters based on mathematical morphology have validly and are widely applied in the field of image filtering.After analyzing the limitations of traditional morphological filters,this paper proposes a new CS morphological filtering model,based on contour-structuring elements,for measuring the morphological extension of connected components in images.A morphological hit operation based on contour- structuring elements is defined.Using the hit operation,the proposed CS morphological filter can extract a noise-area and clear all noises in the noise-area at once.Experimental results indicate that this method has satifactory noise-resistance characteristics and excellent performance for image detail preservation.
Remote sensing monitoring of submerged areas is a direct and effective method to measure flooding and directly reflects the severity of a flood disaster.In this study,the authors monitored the submerged area and dynamic changes of Dongting Lake and Poyang Lake over the middle reaches of Yangtze River with 250 m MODIS data during a flooding event.The results were verified with TM images and showed an accuracy of 94.6%.Further,the results showed that the submerged area of the Poyang Lake in 2010 was larger than other years,while the submerged area of east Dongting Lake and southwest Dongting Lake in 2002 was larger than other years.The longest duration of fluctuation was 5-6 months for the three lakes.Generally,flooding occurred during April and November but most frequently occurred from June to August.Summer is the season with the highest frequency of flood disasters in the middle reaches of the Yangtze River Basin.
This paper presents a streaming data algorithm to execute Delaunay triangulations with large LiDAR point clouds(a billion data points) based on multi-core processor architecture.The algorithm combines divide-and-conquer triangulation with streaming data.A quad-tree structure is used to partition the LiDAR data into subnets adaptively,and schedules triangulation and merging of the subnet data into different processors for load balancing.Parallel computing on multi-core processor architecture makes this algorithm highly efficient with a low memory footprint.
The automation level of classification for remote sensing image need to be improved to satisfy the timeliness and high-precision requirements in disaster emergency monitoring and assessment.But,the artificial selection of typical samples restricts the automatic interpretation of disaster information,a problem particularly acute for the development of business operation systems.This paper implements a totally automatic object-oriented land cover classification system based on automatic sample selection.First,the candidate object samples are acquired by fuzzy clustering.Second,image features and land type features are extracted from imagery and prior knowledge,respectively.Afterward,samples can be selected by applying preset thresholds on these features.Distance metric learning is then used not only for further sample selection,but also for more accurate supervised classification.Zhouqu Debris flow disaster images are computed by this method.Results show that the classification outcomes with samples selected automatically are very close to those samples selected by hand.Our results are more stable and objective than those produced manually.Moreover,it is more convenient to batch process images automatically.
As the floating car data is often with a low precision and frequency,this paper proposes a shortest path based map-matching algorithm for floating car data under spatial and topological constraints,the experimental results on different sampling frequencies data show the algorithm have a promising accuracy.Moreover,the experimental result on floating car data from Wuhan city demonstrate that the algorithm has high reliability which can be used for traffic state extraction and feature analysis.
Simulation is one of the most important methods for researching vehicular ad hoc networks(VANET),but simulation results are usually affected by many parameters.This paper addresses the impact of mobility models on routing protocol simulation in VANET.An Urban Taxi VANET(UT-Vanet) was constructed using floating car data.Based on UT-Vanet,a research program concerning the impact of mobility models was proposed,comparing other mobility models including the random WayPoint(RWP) model,and the IDM_IM and IDM_LC models.Simulation results show that these mobility models have a great impact on simulations of routing protocols in VANET.
In order to solve the problem of low efficiency in high frequency GPS real time data processing when the number of observed satellites exceeds 6-8 while using the DUFCOM method,combining DC algorithm with the advantage of extremely high efficiency,a fast fixed algorithm of GPS single epoch ambiguity resolution by screening and classifying processing for the observed satellites is proposed.The results from the example show that the fast fixed algorithm could avoid the influence of resolution efficiency of double-difference integer ambiguity for the second-level satellites acting on first-level satellites,which improve the efficiency for ambiguity resolution greatly,and achieve the aim to high frequency GPS real time kinematic data processing.
This paper introduces the theory and method for Earth rotation parameter solutions using GPS observations.After several tests and comparisons,more than 80 IGS reference stations distributed globally averagely were to provide observation data with a length of eight hours as a calculation epoch.The ERP time series at the middle point of an epoch with an interval of two hours was obtained by utilizing precise positioning software,GAMIT.The high-frequency time series of earth rotation parameter was initially de-noised by wavelet and then analyzed with fast Fourier spectrum analysis to get the diurnal and semi-diurnal ERP variations.Finally,high-frequency ERP variations with compared with tidal information to determine the excitation mechanism.
Constrained by GPS from CMONOC between 1998 and 2004 in Sichuan-Yunnan region,based on geological division of blocks,we inverse the parameters of active blocks about their present-day crustal movement and deformation by rigid body model and movement and strain model separately.With the comparison of these two models by model identification,it is concluded that it is more proper to describe the properties of crustal movement of active blocks by movement and strain model in Sichuan-Yunnan region.Through the analysis of the movement of active blocks and main faults,the results showed that the characters of crustal movement can be represented by clockwise motion,weakened gradually from west to east and from north to south,and the main feature of crustal deformation is continuous and with gradient shear in Sichuan-Yunnan region.?
Based on the rapid development of CORS system and refined model of centimeter-level quasi-geoid in recent years,this paper implemented the application of the GPS elevation.The technical indicators with low-grade GPS elevation measurement instead of the third and fourth grade leveling are given by theoretical analysis.The result have been verified by the engineering practice.
This paper analyzes the Compass observations through the zero baseline experiments,including the time series of signal to noise ratio(SNR),multipath and the Compass single-difference residuals,which change with the elevation angles.Compared with GPS observations,the Compass signals have some differences.For example,GPS L1 signal is stronger than L2 signal,while the Compass B2 signal is stronger than B1.The Compass carrier phase single-difference residuals changes with the elevation can be used to refinement the elevation-weighting method and SNR-weighting method.The experimental results show that the accuracy of Compass observations are reliable,which can be used in high-precision navigation and positioning,meanwhile,the quality analysis of the Compass observations can provide the basis for building stochastic model.
It is difficult to obtain precise on-board clock short-term stability for current BDS and Galileo common users since there is no public precise clock data.Based on the analysis of single station estimation method of satellite on-board clock stability,this paper proposes a simple method based on smoothed broadcast ephemeris.The principle of this method is discussed,followed by performance evaluation using GPS data.Compared to IGS final clock product and other two methods namely PE and BE which is based on precise ephemeris and broadcast ephemeris respectively,we conclude that the relative estimation error of the proposed method is less than 10% for average time of 1-800 s,which is consistent with the results of PE method and better than BE method.The short-term stability of all the current BDS on-board clocks(until August 2012) are estimated by this method,with the results of 6×10-12(τ=1 s),2×10-12(τ=10 s),5×10-13(τ=100 s),and 2×10-13(τ=1 000 s).
In this paper,the basic theory and mathematical model of two-way satellite-ground time transfer are presented.The error model is deduced and error budgets are analyzed and concluded that the accuracy approaches 0.9 ns.Furthermore,software was developed and actual two-way satellite-ground observations from COMPASS validation testing were processed.The results indicate that the accuracy of satellite-ground time transfer in COMPASS validation phase is 1.0 ns.
This paper gives a brief introduction to the basic principles of ground-based interferometric radar system for dynamic deformation monitoring,and then presents a local relative deformation analysis method for bridge structures considering the characteristics of radar data.Case studies of actual bridge monitoring tests illustrate the feasibility of this method.
The achievements and new progresses in the estimation properties,algorithms,and applications of total least squares in the recent decade are summarized in this paper.Related algorithms and applications of total least squares for processing surveying data are discussed.Moreover,the prospects for algorithms and application of total least squares in geodetic inversion are discussed.
A new expression model for settlement incremental information based on spatial change type,dynamic update operation and graphic data delta is proposed for making up the deficiency of the current expression models,which are based on a single level feature.The relationships among spatial change type,dynamic updating operation and graphic data delta are extracted through the classification of spatial changes.The expression model of settlement incremental information is constructed.The experimental results show that this model is valid and effective.
Network analysis is a fundamental basis for community resources scheduling applications.Discussed in this paper is a multi-criteria-constrained maximal flow problem with route capacity changing over time.Geometric algebra unit based coding is used to obtain a geometric algebra expression for a network diagram.Network connectivity and path search are implemented based on this geometric product.Using independent computation the inner linkage between Boolean operations in geometric algebra computation are exploited,multi- criteria are integrated based on a constrained matrix.A multi-criteria-constrained maximal flow analysis algorithm with changing route capacity is proposed.The algorithm was implemented and validated by maximum flow analysis dealing with pollutant dispersion cases.The results show that,under the constraints imposed by a materials scheduling analysis,the geometric algebra based network algorithm can effectively support multi-dimensional community networks.The algorithm also supports rapid computing and updating of the weight of external constraints as well as changes in the conditions of maximum flow problems.
In this paper,we employ some geostatistics methods including Moran’s I,Moran scatter plot,LISA cluster map and bivariate spatial autocorrelation to study regional spatial autocorrelation and the pattern of aerosol optical depth.Hubei Province is taken as a case study,its spatial autocorrelation degree,the scale of autocorrelation,local spatial agglomeration and spatial coupling features of aerosol optical depth and its impact factors during 2003-2008 were analyzed.The results were as follows: ① The spatial distribution of aerosol optical depth in Hubei Province shows a remarkable level of spatial autocorrelation,the scale of autocorrelation is about 400 km;② There are 2 types of spatial agglomeration area which are high-high and low-low,the areas with high value of aerosol optical depth are distributed mainly over Wuhan Metropolitan and Jianghan plain,and the areas with low value are located in mountain areas of western Hubei Province;③ The degree and pattern of spatial autocorrelation were temporally stable in a comparison between the years 2003-2008;④ Elevation and forest coverage are negatively autocorrelated to aerosol optical depth,the degree of spatial coupling is greater than the population density and total value of industrial output,which are negatively autocorrelated with the aerosol optical depth.
A Virtual Geographical Environment is the expression and simulation of a geographical environment.It not only includes the simulation of the terrain surface information,but includes natural phenomena geospatial and processing simulation.3D terrain visualization is an indispensable part of Virtual Geographic Environment.Due to the nature of two-dimensional features it is difficult,to realistically simulate real three-dimensional volumes based on the traditional "Face model" typically used to build a virtual geographic environment.In this paper,we fully explore the basic concepts of voxels and features,based on the analysis of several common voxel models.We focus on the voxels in a virtual three-dimensional topography of the geographical environment for the simulation of natural phenomena,geographical expression of the process,spatial analysis based on geographic environment and applications.The virtual geographic environment based on "volume model" is dicusssed.
Existing spatial relationship models(topological,directional and metric relations) are generally separated from each other.Furthermore,these models are qualitative descriptions,with low precision.Thus,it is necessary to develop quantified and integrated spatial relationship model.This paper proposes a histogram to represent spatial relationships,the topological,directional,and metric relations between objects,and can quantitatively classify these spatial relationships.Experimental results illustrate the validity of this model.