2009 Vol. 34, No. 5
For the aerial digital frame imagery with large overlap,we propose an automatic transferring point scheme composed of image matching in image-space-free network adjustment-image matching in object-space.Firstly,more accurately exterior orientation elements of images were obtained by image matching in image-space and free network adjustment,which provide approximate values for the image matching in object-space.Secondly,the collinearity geometric constraints were introduced to image matching in object-space,which enables the multi-photo geometrically constrained least squares matching to consider the errors in the image exterior orientation elements flexibly.Finally,the multiple images were simultaneously matched and the reliability and success rate of image are improved.Experimental results show that the proposed method can acquire the image tie points successfully for the images with large overlap,which can meet the demand for automatic point transferring in aerial triangulation,and improve the height accuracy of the photogrammetric points significantly.
Watershed transform is the main tool of mathematical morphology used for image segmentation.This method applied directly to gradient image causes severe over-segmentation.We present an improved watershed transform based on morphological reconstruction and flooding by the analysis of the reasons of over-segmentation produced by watershed and the possible under-segmentation due to watershed-plus-marker.The method proposed here is divided into three steps: ① the alternate sequential filters based on reconstruction are used to reduce some difference inside objects;② the area oriented flooding and depth oriented flooding are applied to gradient image in order to fill some local minima that does not correspond to maximum and minima in original image;③ the processed gradient image is segmented by the watershed transform.The method proposed here can effectively control watershed transform and reduce over-segmentation and under-segmentation.
Image segmentation is prerequisite for object-oriented image analysis.Most image segmentation algorithms need the user to provide parameters to control the quality of the resulting segmentation.Selecting suitable parameters is a challenging task in using such algorithms.We proposed a method of parameters selection for region-growing image segmentation.Information about segmentation parameters was extracted from training sample areas of each class in the image.By multiple-segmentation of the training sample area,a maximum of objective function was found to deduce the suitable parameters for a class.Using the obtained parameters,n(the number of classes) resulting segmentations and subsequent resulting classifications were achieved.Then the n resulting classifications were fused to complete the final image classification.We tested the parameters selection for image segmentation in an object-oriented classification of remote sensing image.
The automatic generation of seamline network is a key problem for the mosaicking of orthoimages over large areas.To solve the problem,we presentd a novel area Voronoi diagram with overlap(AVDO).Based on the AVDO,an auto-generation approach of seamline network was given.The generated seamline network is global based and gives an effective partitioning for the regions of all orthoimages to form effective mosaic polygons,e.g.the area where pixels were contributive to mosaic for each image.The partitioning is unique,seamless and has no redundancy.By this mean in the mosaicking over large areas,we can ensure the flexibility and efficiency of mosaicking,avoid the middle results and error accumulation,and make the mosaicking results have no relation with the order of images.Experiments show that the approach is feasible.
We described the geometric aspects and the mathematical treatment for stereoscopic image pair composed of IKONOS,SPOT4 satellite images and airborne SAR images.First,the rational function model of IKONOS imagery,collinearity equations with line central projection for SPOT4 imagery and F.Leberl model for SAR imagery as well as methods to simultaneously determine or refine individual parameters of these models were briefly reviewed.Firstly,a straight-forward single-step procedure was developed,which follows the philosophy of photogrammetric bundle adjustment techniques.Secondly,the combined stereo location model with multi-sensor remote sensing images was constructed from these geometry models.The stereo location is the procedure of determining the ground coordinates of a ground point from a pair of corresponding homologue points in the combined stereoscopic pair.Experimental results show that it is feasible to realize combined stereo location with multi-sensor images.The accuracy of combined stereo location is close to classical stereoscopic pairs in some conditions.
Based on the classical FCM clustering,the spatial fuzzy membership about a pixel and regions is defined and constrained into the classical partition matrix.Tabu search was introduced to overcome the locality and the sensitiveness of the initial condition of FCM clustering.A FCM algorithm based on Tabu search is parallelized to reduce the communication complexity of image segmentation and to improve the overall performance of the scheme,which achieves a satisfied linear speedup.The experimental results show the efficiency of the proposed algorithm in decreasing clustering iterations and increasing classified precision.
A new hierarchical Markov model in wavelet domain was proposed.In this model,the Gauss Markov random field(GMRF) was used to model the distribution of wavelet coefficient vectors to describe the relationship of observed features on each scale,and the cooperation of interscale casual.Innnerscale non-casual Markov Random Fields was exploited to model the label field priori probability.Based on the Bayesian rules,a new textured image segmentation algorithm was proposed employing multi-objective problem solving technique in this new hierarchical model.Experiments with synthetic texture images and remote sensing images were carried out.The results show the abilities of the proposed method to reduce segmentation error rate.
Error diffusion algorithm is one of the commonly used image hafltoning methods.The quality of the halftoning image is directly affected by the different image interpolation methods.We mainly analyzed the quality of error diffusion halftoning images using the nearest neighbor interpolation,bilinear interpolation,and bicubic interpolation.The image quality method based on human visual model was used to evaluate the qualities of the halftoning images.Experiment shows that the image of error diffusion halftoning using the nearest neighbor interpolation is the best.
A color image segmentation method was presented for Virtual Chinese Human Project.Using this policy,the color slice images of refrigerant body can be segmented in series automatically.This new method was mainly based on region-growing arithmetic.A combination of color and texture was performed for comparability evaluation.Window-median arithmetic was applied to avoid noisy seed point.A corrosion algorithm was implemented to create target seed points automatically.An adaptive algorithm was adopted to set threshold.And the multithreading frame for concurrent computation was introduced in the entire segmentation procedure.This method supermatic the primary organ of Virtual Human was segmented fleetly and serially.The organ's figures after segmentation were clear and accurate.The models after 3D reconstruction were visual and vivid,as well.
The correspondence analysis(CA) method,a multivariate technique widely used in ecology,is relatively new in remote sensing.In the CA differencing method,bi-temporal images are transformed into component space,and individual component image differencing can be performed to detect potential changes,somehow similar to principal component analysis. The advantage of the CA method is that more variance of the original data can be captured in the first component than in the PCA method.However,these techniques are all performed on a pixel by pixel basis,becoming unsatisfactory in some circumstances due to higher spectral heterogeneity in imagery of high spatial resolution.This problem can be alleviated by the object-based strategy,which segments the image into regions of relative homogeneity,which are,in turn,used as the basic units for data analysis.We proposed an object-based CA approach for change detection,whose performance was compared with those of pixel-based PCA and CA.Experimental results show that the object-based CA method produces the best accuracy in change detection.
We introduced the baseband loop synchronization design of BD satellite navigation system.We build a model of carrier and code tracking loop system,and analyze the effects of different loop parameters producing on system's performance.Considering of the symbol synchronizating,we adopted Farrow structure to interpolate the data.The simulating results show that the loop parameters do have influence on the rising time and the static error of the tracking loop.The whole tracking loop system can be good at demodulating and de-spreading the original data.
In order to restrict the models error of fitting gravimetric quasi-geoid to GPS/leveling quasi-geoid,the new method of fitting two kind quasi-geoid using Bayesian regulation BP neural network was proposed.Using the gravimetric quasi-geoid and GPS/leveling data in a certain area,the new method was compared with polynomial surface fitting method.In the case with biggish area and anomalous difference between two kind of quasi-geoid,Bayesian regulation BP neural network could reduce the erros of models,and Bayesian regulation arithmetic could improve the structure of network by restricting weights to produce a smoother network response.The experimental result shows that the new method can improve the inner and outer precisions of fitting two kinds of quasi-geoid clearly.
We introduced the exploratory spatial data analysis(ESDA) and Kriging theory,and proposed an improved spatial variability-oriented Kriging interpolating model for GPS height transformation.Five kinds of typical modeling scheme were put forward through combining Kriging interpolation factors with the ideas of step-by-step comparison.The error results shows that the No.4 scheme is the best one,which is called ESDA-K.We did an experiment with ESDA-K,neural network to simulate GPS height anomaly.Experimental results show that the simulating accuracy of ESDA-K is more excellent than that of traditional methods.The ESDA-K model's precision is stable,and the model resolves the questions that traditional methods must face,such as local undulation,over-learning,falling into regional maximum easily etc.
A new algorithm for computing geodetic coordinates from Cartesian coordinateswais is derived by solving a modification of Levin's quartic equation with Newton's iteration from two different initial values.Experiments with two given initial values show that the new algorithm is sufficiently precise,fast,and stable.
Double-star navigation position system adopts the mechanism of double-star position,and the basic position principle is three-sphere intersection measurement.Two observations are related to the distances between the two satellites and the survey station,and the third observation is related to the survey station's height,which is the critical factor that determines the system's position accuracy.Hence,we introduced the concept of similar ellipsoid,and deduced the accurate model of the third observation.Then,we established the position algorithm.The simulating experimental results show that this algorithm has a quick convergence rate and high accuracy of position.
High precision GPS horizontal velocities and gravity data were used to estimate the crustal motion and fault slip rate in Sichuan-Yunnan region(96°-106°E,20°-36°N).Due to the influence of India-Tibet block's east-northward extrusion,and the gravitational buoyancy force associated with the sharp topographic gradient across the region,the first-order features of crustal deformation are the prominent clockwise rotation around East Himalaya Syntax(EHS),and leads left-slip motion along the Xianshuihe-Xiaojiang fault system(east boundary) and right-slip motion along the Jinshajiang-Honghe fault system(west boundary).The Xianshuihe-Anninghe-Zemuhe-Xaiojiang fault system is the most active left-lateral fault in the region,with a rate of 12.1±0.6 mm/a,9.0±1.2 mm/a,6.4±1.0 mm/a,6.0±1.2 mm/a and 9.0±1.2 mm/a respectively.The Longmenshan fault system is with a rate of 2.6±1.1 mm/a right-lateral slip,and 1.3±1.2 mm/a extrude.As a result,the pattern of crustal deformation in the region supports the continuous deformation hypothesis.
The precision of GEO orbit plays an important role in regional navigation system when the hybrid constellation of GEO/IGSO/MEO was involved.Determintion of the orbit of GEO during orbital maneuver is important.A first order Markov stochastic process was used to construct the thrust acceleration model,and the correlative parameter estimation method was developed.Based on this algorithm,the TWTT(TWO-WAY TIME TRANFER) measurement was used to determinate the GEO maneuver orbit.Eperimental results show that this algorithm is remarkable to fit the maneuver orbit.
The essential component and the characteristic of GNSS software receiver were described according the principle of GNSS software receiver.The key modules,which include the signal processing module,the signal acquiring and tracking module,and the navigation ephemeris resolution module,were introduced emphatically.In order to test the effectiveness of these modules,actual collection results of signal acquisition,tracking and positioning precision were performed.The results demonstrate that the proposed algorithms of signal acquiring,tracking,and navigation resolution can be carried out correctly and conveniently,and the single point static positioning can be realized.The result precision can reach 10 m when GDOP value is less than 4.
We summarized the basic processing steps of pulsar timing model correction.The calculating methods of Einstein,Shapiro,Roemer,dispersion delay,and the the binary pulsar models in common use were given.Order of timing delay correction magnitudes were calculated using milisecond binary pulsar data.The pulsar was called J0437-4715.The data Is provided by Australia Telescope National Facility.The influence of calculating results using different binary pulsar models and planetary ephemeris were analyzed.
We presentd a χ2 method to detect the atomic clock anomaly of navigation satellites by making atomic clock bias twice difference and biulding the balanced time serial.Experiments show that double difference method is effective for the phase anomaly and not effective for frequency anomaly,and χ2 method based balanced time serial is effective for both phase anomaly and frequency anomaly.
We pointd out several key issues,which should be well resolved before the precision analysis of GPS broadcast ephemeris.Statistical analysis of both the orbital and clock accuracy were presented with a test of a period of two weeks,mainly from the aspect of satellite itself.The orbital accuracy reached 2-3 m,and the clock precision reached about 10 nanoseconds.In addition,The precision varies from satellites to satellites.Satellites of Block IIR were superior to that of Block IIA,while rubidium clocks were better than cesium clocks,especially in the acpect of stability.
The direction of beam width-effect has a significant influence in bathymetry.The slope angle of seabed is the key problem of the beam width-effect correction in bathymetry.Based on the theoretic analysis of beam-width effect,a slope angle calculation model was proposed,which consideres the direction of survey lines.The results of numeric simulation in different headings show that the model is valid and could get the most accurate slope angle when the survey line heading is the depth gradient direction.
The problem about multi-representations of spatial data is one of the hot topics in modern GIS.We pointed out that all kinds of published solutions could be concluded as three kinds of type techniques: explicit storage of multi-scale vector data,multi-scale spatial index,multi-scale vector data storage structure.Because there is more fertile soil to seed multi-scale spatial index method,we expatiated six kinds of multi-scale spatial index method,such as Reactive Tree,GAP-tree,Multi-Scale Hilbert R-tree,Multiple R-tree,and followed their development in recent years.According to our research experiments,we discussed their advantages and disadvantages,and provided some academic bases for their chosen and applications.Finally,we drawn some conclusions to guide the research on multi-scale spatial methods,and proposed the further research on multi-dimension index and optimized index.
By means of spatial relation analysis,we discussed the limitations of current proximity models among area features,and formed a cognitive framework that external and internal distance between central points,tendency length of boundaries are the three key-factors of this issue.We proposed the lateral,the depth and the whole proximity,and set up the quantitative models based on partitioning central line by boundaries and constrained Delaunay triangular network.Finally we defined the data structures and constrcting methods of proximity for area features layer.
The space-based information system,(SBIS) is an important aerospace equipment to obtain the information superiority in the battlefield with a dynamic network of spatial-temporal feature.A modeling for the SBIS network topology was proposed,which is on the basis of the modeling of nodes and links,using static and dynamic modeling to describe the SBIS network topology regularity in the operation.The static model describes the topologic structure at a given break.In the proposed dynamic model,time break and timeslice were defined,and an algorithm of timeslice dividing was presented.Finally,a simulation sample with 12 nodes was built.The simulated results indicate the topology models built are reasonable that can reflect the regularity of the SBIS network in a certain extent.
We proposed a new method for sharing symbols between different GIS platforms by virtual machine technology.The symbol system consists of three parts: symbol virtual machine,virtual machine articulated plug-in,and symbol design tools.Symbol virtual machine is a software simulated machine,which has virtual hardware and corresponding instruction system.Symbol virtual machine is independent of any specific GIS platform,responsible for dealing with GIS platform-independent function of symbolization.Symbol virtual machine articulated plug-in connects the symbol virtual machine with the GIS platform.Hence,the GIS platform can call symbol virtual machine to do symbolization.Symbol design tools are used in the production of symbols.When producing symbols,we only target at symbol virtual machine,not target at specific GIS platform.Thus,the symbols can be shared in all GIS platforms articulated symbol virtual machine.
Based on 3D cell grids,we proposed a new K-nearest neighbors search algorithm.The point cloud was divided twice and distributed to 3D cell grids,then ultimate space,internal space,external space were decided for each grid.With the help of each point's sphere space,K-nearest neighbors of the point can be found quickly.Compared with the existed methods,the proposed algorithm has more efficient performance.
We proposed a new watermarking algorithm for the digital grid map based on integer wavelets transformation.The digital grid map was decomposed by integer wavelets transformation firstly.Then the watermark was embedded into low frequency section and high frequency section respectively.The watermark was adaptively embedded into low frequency section according to the luminance,and embedded into part of high frequency section as well.Which is selected by the characteristics of human visual systems and the intrinsic relations of wavelet coefficients.The experiments show that the proposed algorithm is robust against various vicious attacks such as JPEG compression,sharpening,cutting.
The loading of spatial data servers can not be dynamically adjusted according to the characteristics of spatial data using current load-balancing methods in large-scale spatial data Web services.Aiming to grid data,we proposed a spatial-data-content-based dynamic loading balancing method,which can dynamically divide spatial data servers into several server groups according to geographical position relativity and accessing heat of spatial data.This method improves the cache utilization on every spatial data server.
C-means clustering is the first objective function clustering algorithm.Our purpose is to set forth a C-means algorithm based on data field,which can be widely applied to the classification and hierarchy in the field of spatial data mining and knowledge discovery.The basic idea of this method is to find the initial cluster centers based on the introduction of data field.Priority should find potential heart k according to the potential field distribution function,then select sample data being close on the potential heart.Finally,we can achieve C-means clustering algorithm.We presentd the implementation steps of the method,and verified the feasible and scientific significance by examples.