2010 Vol. 35, No. 11
An improved and rapid arithmetic for multi-circle detection using randomized hough transformation(RHT) based on edge points' cluster analysis,is put forward.The result shows that this method can rapidly and accurately extract multi-circle objects on complex images,which has tremendous value in use especially when the objects are separate and own larger shape.Meanwhile,this arithmetic has a better ability of resisting noises.
This paper presents a novel approach of detecting special objects from high resolution satellite imagery.In this approach,a bilateral filtering is used to reduce the noise firstly,and a morphology approach is proposed for ROI extraction and feature enhancement.A detection operator based on the derivative of the double exponential(DODE) is developed to detect enhanced objects.The operator is designed according to special objects.The experiments which include vehicle detection and ship detection from Google Earth images are discussed.The experimental results show that proposed approach is effective and feasible.
We propose a supervised method for texture segmentation using ant colony optimization algorithm(ACO) to extract prominent Gabor features.Firstly,training samples are randomly selected from the texture images and exposed to the Gabor filters transform.Secondly,a feature subset corresponding to each sample can be selected by minimizing the distance summation between Gabor texture vectors and the different texture classes based on ACO.It aims to determine the central frequency and orientation of Gabor filters globally.The objective makes the frequency responses between different textures well separated.Finally,K-means algorithm is chosen to segment the test images based on reduced features.The experiments on two synthetical textured images show the promising results using the proposed method.
A dynamic shadow detection algorithm for aerial images captured by micro unmanned helicopter is presented.To meet the two requirements of the application platform which are high robustness for various environment and high adaptation to moving background,two independent features are extracted based on the analysis of shadow feature set.A modified Co-EM strategy is applied to shadow detection for separate frames.The experimental results show that this algorithm can not only be applied on real micro unmanned helicopter platform,but also provide better detecting results than the regular method which uses color and region features.
We analyze the spectrum influence between Radon transform and Log-polar transform when rotation effect is eliminated.The average retrieval performance of wavelet and NSCT with different retrieval parameters is also studied.Based on which,we design a multi-scale and multi-orientation texture transform spectrum,as well as rotation-invariant feature vector and its measurement criteria.Then a new two-level rotation-invariant texture retrieval algorithm based on no-parameter statistic features is proposed.The experimental results on Brodatz image database show that our algorithm is appropriate for main orientation capturing and detail information description.The combination of this two-level progressive retrieval strategy and multi-scale analysis method can effectively improve retrieval efficiency,compared with traditional algorithms,and ensure a high precision as well.
An automatic method of simulating low resolution panchromatic image in IKONOS image fusion was described.Firstly,support vector machine(SVM) was used to separate pixels containing low frequency information from those containing high frequency information,which was not suitable to be included in regression coefficients estimating.Secondly,improved Bucket technique was adopted to generate a subset of observations evenly distributed.Finally,low resolution panchromatic image was simulated with parameters achieved by linear regression,and integrated into the Gram-Schmidt spectral sharpening method.Validating experiments were carried out on two datasets of IKONOS panchromatic and multispectral images,visual and quantitative quality judgments show that the proposed method can select evenly distributed pixel observations with low frequency information automatically,which proves its high efficiency,and that the resultant images have less spectral distortion than traditional Gram-Schmidt method does.
We propose an image segmentation method based on cellular automata and fuzzy C-means(FCM).The method defines the objective function based on FCM and takes gray level as the object to calculate the minimum objective function by application of one-dimensional cellular automata's self-iteration,realizing the preliminary image segmentation;on that basis,it revises original segmentation results with two-dimensional spatial information of neighborhood.This method can be used for image segmentation not only in grayscale images but also in RS image.
Haze is an important influence factor of visible light RS data's obtaining and using.Based on dark channel prior and haze image model,we study the dehaze technology from a single image.Aim at the problem of dark channel prior not work for special region of city image and traditional medium transmission computing method's disadvantage: complex,time-consuming,making use of RS image's characteristics: haze distributes relatively more uniform and depth of field change not acutely,proposes a new quick computing method on the premise of not markedly reduce dehaze effect,to realize single image fast dehaze processing.Using multiple frame images to make system experiment,and do quantitative analysis to dehaze result images' definition.
Because of the restriction of gray registration strategies on improving the accuracy,the SIFT was combined with data snooping algorithms which came from the photogrammetry.The wrong registration feature points were processed as the gross error in the quadratic polynomial model and adjusted by the error snooping.The experimental results proves the reliability of this method.
The chirp scaling(CS) algorithm for forward-looking bistatic synthetic aperture radar is proposed.Firstly,the imaging theory of forward-looking Bi-SAR is studied and its signal model is either constructed.Then the forward-looking Bi-SAR is degraded into the fixed receiver Bi-SAR by time domain linear range migration compensation,and the azimuth Doppler modulation rate change of the degenerated signals are analyzed;Based on these,the Bi-SAR chirp scaling algorithm is brought forward with the new match functions.Finally,the computer simulation verifies the validity of this approach.
A novel SVD based gray-level digital watermarking algorithm is presented.In this algorithm,data from high bit to low bit of decomposed bit planes of the gray-level watermark image are embedded respectively into corresponding singular values from large to small obtained by SVD of sub blocks of the original image.And that the revision for resistance against brightness-and-contrast adjustment is adopted makes that this algorithm has the robustness against brightness-and-contrast adjustment attack.The experimental results show that this algorithm has fine transparency of the embedded watermark and has extremely strong robustness against brightness-and-contrast adjustment attack,and it has as well strong robustness against other common image-processing attacks such as image smoothing,noise adding,color dithering,and JPEG-compression.
A encryption approach of Montgomery-Form ECC for remote sensing images is proposed,combining with faster computing speed of scalar multiplication on Montgomery-Form ECC compared with Weierstrass-Form ECC,mutual authentication of ECDH key exchange system based on certification authority respectively.The simulation analysis shows that the approach improves computing speed for mass data of remote sensing images;strengthens security of plaintext embedding;achieves mutual authentication between sender and recipient,and works for encryption and decryption for mass data of remote sensing more efficiently.
We analyzed the current status of remote sense image sharing technology.The SOA-based sharing architecture was proposed to solve the problem of isolated information islands.A metadata schema was defined based on OGC specifications to supports search,evaluation and integration of data services via computer application.Metadata services and image data services were designed and implemented.The result shows that the SOA-based sharing architecture is technically feasible to integrate distributed heterogeneous image data resources.
We propose a method with the following procedure.Firstly,we correct the elevation of some rough ground by surface fitting.Secondly,we track the contour point by building triangle mesh and eliminate burrs by weighting quadratic polynomial.Finally,we interpolate to get the smooth contour by spline function with contourinterval.The experimental results show that our method can generate uncrossed and smooth contour in less loss of precision.
In this paper,a new approach based on whole object is pursued to compute the binary topological relationship between unchanged contour line and changed contour line,in which FL points are introduced,intersection and difference operators are selected from set operators to distinguish the topological relations between neighboring spatial line objects;three types of topological invariants are used for the computational results of set operations: contents,dimension and connectivity-number.Then a computational method of contour line fusion is designed and realized.The proposed approach of contour line fusion is operational;it has been used in the automatization of contour line fusion successfully.
In this paper,zero-difference kinematic precise orbit determnation method is adapted for GRACE by using TriP software.Considering the effect of data quality,some observational data of GRACE were solved to obtain appropnate catoff elevation and weighting stategy.Using the optimal strategy,we processed four months observational data of GRACE and compared the orbits with the products provided by JPL.The results indicate that the accuracy of orbit determnation of TriP reaches 5-6 cm,which is at the same level of national and international studies.
The systematic errors of GPS data from GRACE satellite are analyzed,and the numerical results indicate that the multipath error is not significant while the occultation antenna is turned off.But when the occultation antenna is turned on,in the aft-looking hemisphere,there are clearly multipath-like systematic error in the code observations which is caused by the cross-talk between the prime POD and occultation antenna string,but the analysis indicate that the cross-talk systematic error in the carrier phase observations can be ignored.So in the orbit determination processing,the weight of the code observations are adjusting to reduce its impact on the POD results.The numerical results show that when the STD of the P3 observation is greater than 4.0 m,the POD results can achieve the best precision,which is 3.9 cm,4.0 cm,and 3.4 cm in X,Y,Z respectively.
Different from traditional single-step algorithm,focus on satellite orbit integration from acceleration to position using new 12 steps Runge-kutta 2 orders algorithm.This algorithm is introduced in artificial satellite orbit application,and its integration results are studied and compared with other algorithms'.The results indicate that for integration precision and computation stability this algorithm is advanced,but it need more time consuming than others at same integration step.Synthetic consideration of the algorithm's stability whose integration error does not obviously increase with step length's augment,we can increase step length and decrease computation time to fulfill the needs of high-orbit satellite's orbit prediction and precise orbit determination.
We analyzed and comparised the results of precise point positioning(PPP) using 5 min and 30 s intervals precise satellite clock offset provided by IGS,analyzed the influence of different kinds of precise satellite clock offset to the PPP result,with emphasis on initialization convergence time and positioning accuracy.The experimental results show that the accuracy results of static and kinematic PPP using 30 s precise satellite clock offset were better than those using 5 min precise satellite clock offset interpolation into the 30 s satellite clock offset.
Based on satellite-bone GNSS,the precision orbit determination(POD) for low earth orbit satellites is becoming one of hot spots in the realm of geodesy and geomatics.It is also one of means of POD for the domestic earth-oriented satellites.In this paper,accuracy evaluation methods used currently are discussed in details as well as their advantages and disadvantages.The experimental results are carried out so as to verify the related methods on different conditions and to check their validity and limitations.
The method of ridge estimation to treat ill-posed and weighted total least squares adjustment(TLSA) is put forward.The solution equation and evaluating precision by mean square error are deduced.The model parameter discernable matrix in ill-posed and weighted TLSA is defined,then the meaning and effect of the ridge parameter is discussed.For the case that the ridge parameter is difficult to determine in ridge estimation to treat ill-posed and weighted TLSA,the L-curve method,generalized cross-validation(GCV) and the ridge mark method are introduced and given.At last,through some cases,the comparisons are carried out among the L-curve method,GCV and the ridge mark method,also the results of common least squares(LS) and total least squares are given,and some conclusions are obtained.
Total least squares(TLS) approach aims at estimating a matrix of parameters from a linear model when there are errors in both the observation vector L and the data matrix A.The authors derived an iterative algorithm to solve the TLS problem by using the principle of indirect adjustment.Compared with the method based on singular-value decomposition,the iterative algorithm coincides with the SVD algorithm.The calculated example has proved that the iterative algorithm is validity and rationality.
In order to realize the precise tracking of GPS code in the environment of satellite motion,discriminators gain conception is given and the gain forms of the usual discriminator are derived.Then the deduction is given that the larger the discriminator gain is,the higher the pull in speed is and the lower the steady tracking precision is.The effect of correlator space,loop bandwidth and the correlator gain on precise tracking are discussed.The analysis is given that correlator space decrease deduces the tracking error more than dynamic decrease.The correctness of the deduction and the analysis are validated by the Matlab simulation platform.At last,a more reasonable loop design scheme is proposed based on narrow correlator and dot production discriminator.
The concept of edge variation factor is firstly defined based upon the distance variation among the entities in the spatial neighborhood.An approach is presented to construct the minimum spanning tree-like(MST-L for short).Further,a MST-L based spatial clustering algorithm(MSTLSC for short) is developed.Two tests are implemented to demonstrate that the MSTLSC algorithm very robust and suitable to find the clusters with arbitrary shape,especially the algorithm has good adaptive characteristic.A comparative test is made to prove the MSTLSC algorithm better than classic DBSCAN algorithm.
These existing direction models are difficult to consider the fuzziness of direction concept,the level of direction relationship,the uncertainty of spatial data,the local character and global character of spatial direction relation.Characters of direction relation and extended 8-direction asymmetry fuzzy model were introduced at first,the extended 8-direction asymmetry fuzzy model was developed after considering the location error of reference point based on interval type-2 fuzzy set,and the primary membership function and uncertainty of membership was discussed,the FOU of fuzzy directions was analyzed too.Under this model,the reasoning method was built based on IT2 FLS,the KM defuzzifier method was used in the process,and two variables were used to measure the uncertainty of direction center.This method takes the location error of reference point,cognitive habit and percentage of target in each direction tile into account.
We apply shape-number method to represent polygon shape and by least editing distance computes the shape similarity.The simplification of building polygon is conducted by replacing the original polygon with the matched template object.The method based on shape matching is an attempt of application spatial recognition in map generalization.The experimental results show that the matching similarity based on the shape-number calculating is consistent with human cognition able to be practically used in building generalization.
This paper uses HASM method to construct DEM from stream lines and scattered points.A case study on the typical landform of plateau of China,Kongtong gully of Dongzhi Plateau,Qingyang,Gansu Province,is presented.The original data is manually vecterized from scanned maps 1∶10 000,includes scattered points,stream lines,contour lines,precipitous cliff lines of a region of area about 4×5 square km.For simplicity this paper uses only the former two of them.Compared with the result from stream-line-constrained TIN and hydrologically correct DEM construction method thin plate spline(TPS),through visual inspection,HASM gets a more desirable DEM and more reasonable integration of information of the terrain features.Some challenges and future research about HASM is also given.
In this paper,the theories and overall frameworks of land market prosperity analysis were built,the prosperity index method was introduced and applied to the research of land market prosperity analysis in China from 1994 to 2006.The results have indicated that the operation status of land market in our country has been generally good since 1994.Comparatively speaking,the land market on the former stage(from 1994 to 2002) operated very stably,except that it has been depression in 1999,in others years the land market has in the normal state.On the latter stage(2003-2006) the land market fluctuated very obviously,it has been heat in 2003,2005,and cool in 2004,and has been normal until 2006.All of these mentioned above are the driving results of a series of macroeconomic control policies,including many land management policies and economic development measures,etc.
Using MODIS-derived aerosol data,temporal variation and spatial distribution of aerosol optical depth(AOD) during 2003-2008 over Hubei province was investigated based on GIS.SOM(self-organizing maps) and linear models were used to analyze the relationships between AOD and population density,per capita gross industrial output value,forest coverage.The results were as follows: high AOD values were observed in central south areas with lower elevation,larger population density,higher per capita gross industrial output value,lower forest coverage,while low values,in western,northeast and southeast,were contrary.Distribution of seasonal average AOD and η(fine-mode fraction) vary from each other.Monthly average AOD rises substantially in winter-spring season and drops sharply in spring-winter season.Two models were both showing that AOD distribution is positively related with population density,per capita gross industrial output value,and negatively with forest coverage.