2017 Vol. 42, No. 7
Land surface temperature (LST) is one of the important biophysical variables affecting the exchange of water and energy between land-surface and atmosphere, and it is significant to retrieve LST accurately. The mono-window algorithm is more applied in Landsat TM 6 data, including Jiménez-Muñoz mono-window algorithm (JM_SC) and Qin Zhihao mono-window algorithm (Q_SC). There are a lot of changes for Landsat8 thermal infrared sensor (TIRS) compared with Landsat TM6. Thus a mono-window algorithm for Landsat 8 data (TIRS10_SC) was proposed first, and then some comparison and analysis of three mono-window algorithms were conducted in this paper. The results show that:(1) The TIRS10_SC algorithm is closely integrated with the characteristics of Landsat8 TIRS sensor and it performs well to retrieve LST of different land-cover types based on the retrieval of atmospheric transmittance and land-surface emissivity. (2) Through the comparative analysis, it is found that the retrieval accuracy with Q_SC and TIRS10_SC is higher than JM_SC algorithm. (3) The retrieval results of homogeneous underlying surfaces such as bare soil land and cement surfaces are more accurate than vegetated surfaces. For bare soil land and cement surfaces, the average error of TIRS10_SC and Q_SC algorithm is 0.60℃, and that of JM_SC is 1.01℃; for vegetated surfaces, the average error of TIRS10_SC and Q_SC algorithm is 1.48℃, and that of JM_SC is 1.26℃.In order to improve the LST retrieval accuracy of vegetated surfaces in urban areas, the land-surface emissivity characteristics of vegetated surfaces need to be quantified more accurately.
According to the essential feature of object-oriented image segmentation method, this paper explores a minimum span tree (MST) based image segmentation method. We define an edge weight based optimal criterion (merging predicate) which based on statistical learning theory (SLT), a scale control parameter is used to control the segmentation scale. Experiments based on the high resolution UAV images show that the proposed merging predicate can keep the integrity of the objects and do well on preventing over segmentation. It also proves its efficiency in segmenting the rich texture images while can get good boundary of the object.
We propose to use fuzzy relation based co-occurrence kernel for classification of high-resolution aerial images. By analyzing the characteristics of aerial images, it points out that the imagery does not have an absolute reference frame. For this reason, it uses a local descriptor called MROGH which is inherently rotation invariant to extract low-level features of aerial images. It then uses fuzzy relation based spatial co-occurrence kernel to build the holistic representation of aerial images. Experiments results on publicly available aerial scene imagery dataset show that our method gets a better classification result. In addition, we make a consistent comparative analysis of different classification frameworks based on aerial image dataset.
The traditional pixel-wised classification methods for hyperspectral image (HIS) only consider spectral information while ignoring the spatial information, resulting in a big limit of classification performance. Clustering which could assemble pixels similar in spectral features into spatial adjacent clusters, thus effectively express similarity and spatial correlation of adjacent pixels. In order to take full advantages of spatial correlation, this paper explore a spectral-spatial classification method for HSI merged with clustering and context. Firstly, under condition of different feature extraction(MNF, ICA and PCA), different clustering methods(k-means, ISODATA and FCM) are used in hidden markov random field to obtain optimized segmentation map containing context features; secondly, the regions in the segmentation map are labeled by using a four-connected neighborhood labeling method to generate image objects, and a majority voting method is used to classify the objects based on the initial classification map derived from support vector machine (SVM) optimized by particle swarm optimization (PSO). Finally, a Chamfer neighborhood filtering technique is used to regularize the classification map, which partially reduces the noise. This method utilizing spatial information from clustering and introducing context features from HMRF takes advantage of supervised classification and unsupervised classification to gain noise reduction, high-accuracy and high homogeneity, which makes up for the inadequacy of the classification based only on spectral information. Experiment on ROSIS data set and AVIRIS data set respectively illustrate that the method can obtain better performance in terms of classification. The overall accuracy of ROSIS data set reaches to 98.53%, 5.01% higher than that obtained by SVM. Meanwhile the overall accuracy of AVIRIS data set climbs to 91.97%, 7.01% higher than SVM result. We also find that different feature extraction and different clustering will influence the spectral-spatial method using HMRF with edge-protection.
The binary image method can automate extraction of tidal datum referenced shorelines from light detection and ranging (LiDAR) data. With this method, spatially detailed and continuous shorelines can be derived. In this paper, an improved binary image method is developed in order to promote efficiency. Firstly, data preprocess is applied to the LiDAR data, including coordinate transformation and noise reduction, with the exception that no filter is used to generate an LiDAR DEM. Secondly, the LiDAR data is segmented into a binary image by intersection with the MHWS (mean high water springs) datum surface, whose height calculated by the tide gauge data or tidal model. Thirdly, the tidal datum referenced shorelines from the LiDAR data are extracted by a chain of image processing procedures, including object recognition, bogus-objects deletion, and edge detection. Finally, the accuracy of shorelines is evaluated by the comparison of LiDAR-derived shorelines with the ground survey data. An application shows that the improved method is more efficient and more reliably than the binary image method.
HUT model (Helsinki university of technology snow emission model) is a semi-empirical radiative transfer model of passive microwave remote sensing of snow, and it considered the impact of soil, forests and atmosphere on brightness temperature receiving by the sensor. HUT model was able to simulate brightness temperature receiving by the sensor. This paper validated HUT model based on MWRI data mounted on FY-3B satellite. The results showed that there was a big difference between brightness temperature of HUT model simulation and brightness temperature of MWRI for 18.7 GHz horizontal polarization and 36.5 GHz horizontal polarization. So this paper improved the extinction coefficient formula. The model that used the improved extinction coefficient formula calculating extinction coefficient was called improved HUT model (IMPHUT model). The validation result of IMPHUT model indicated that:the bias of IMPHUT model for 18.7 GHz horizontal polarization and 36.5 GHz horizontal polarization were -0.91 K and -4.19 K respectively. However the bias of HUT model for 18.7 GHz horizontal polarization and 36.5 GHz horizontal polarization were 14.03 K and -16.33 K respectively. Simulation accuracy of IMPHUT model was greatly improved. Finally, genetic algorithm was used to inverse snow depth on January 20, 2013. The result showed that bias of inversion snow depth based on IMPHUT model was -6.79 cm. The inversion snow depth based on IMPHUT model which was superior to HUT model and Chang algorithm was in good agreement with the measured snow depth.
To overcome the defects of existing algorithms that the target information and edge details are easily lost and that fusion image contrast is low, a novel fusion method that combines region feature and multi-scale transform for thermal infrared and visible images is proposed in this paper. Firstly, the source infrared and visible images are segmented based on adaptive pulse coupled neural network (PCNN) and two-dimension Renyi entropy, and a joint segmentation map can be acquired through region joint operation. Then the original images are multi-scale and multi-directional decomposed by nonsubsampled contourlet transform (NSCT). After that, the fusion rules are designed based on region feature difference in NSCT domain. Finally, the fusion image is reconstructed by NSCT inverse transform. Experimental results show the proposed method can effectively fuse infrared target feature, preserve the background information as much as possible, and obtain good contrast. The proposed method is superior to the traditional methods in terms of both subjective evaluation and objective evaluation.
The two-line CCD stereo camera carried by the Chang E-2 (CE-2) satellite has successfully captured the global lunar imagery at 7 m resolution by the linear push-broom imaging manner. Compared with the image of Chang E-1(CE-1), the spatial resolution of CE-2 is greatly improved, which enables the more detailed presentation of the lunar surface features and topography, and consequently makes the production of high-resolution lunar DEM more complex and difficult. The existing methods based on the Chang E image can hardly meet the application requirements. In this paper, based on the previous studies on the characteristic of CE-2 image, we proposed methods of image matching and the Digital Elevation Model (DEM) extraction. The working flow started with the generation of the approximate epipolar images based on the RPC model and the construction of the pyramid images. By matching with the pyramid hierarchical, a pixel-wise matching of the original image was realized. We then calculated the three-dimension coordinates on the lunar surface of the matching points, and generated the DEM using the inverse distance interpolation method. Finally, the radial basis function interpolation was employed to fill the terrain hole caused by the image shadows. The results show that the lunar DEM generated by our method can accurately depict the most detailed and complete lunar surface topography, and perform better in the production of the global Lunar DEM products of CE-2 mission.
Spatio-temporal abnormal cluster pattern is an important spatial point pattern. The pattern results can reflect the distribution and evolution of spatio-temporal events timely and accurately. Early researches has verified the scan statistic based clustering methods are very effective in detection spatial and spatio-temporal abnormal cluster pattern. However, due to the fixed shape of scan window, traditional scan statistic based clustering methods have limitation on obtaining exact shape and size of cluster. This paper proposed an improved irregularly shaped spatio-temporal abnormal cluster pattern mining algorithm stAntScan. The algorithm constructs the spatio-temporal neighborhood matrix by a newly defined 26 directions spatio-temporal neighbor cells. Then the algorithm improves the ant colony optimization based method to fit for spatio-temporal scanning on three-dimensional large data set. In the end, the Monte Carlo simulation method is used to test the significance of clusters. Experimental results on both simulated data and real Weibo check-in data have testified the efficiency and accuracy of stAntScan on irregularly shaped spatio-temporal abnormal cluster pattern mining. And compared with the classical SaTScan, it gets much better results in finding exact shape and size of clusters.
In order to solve the difficulty that the existing algorithms can't give attention to both the capability of time and space when constructing Delaunay triangulation with massive LiDAR points cloud, a cutting block Delaunay triangulation algorithm using streaming computation is presented. DeWall (Delaunay wall) is constructed firstly to cut an independent point cloud data block with specific size and shape, which is suitable for the divide-and-conquer algorithm and can avoid deep recursion and memory overflow. Then the cutting block is triangulated with a divide-and-conquer algorithm, and an algorithm is proposed to delete the wrong triangles on the boundary of the cutting block triangulation. The process described above is repeated to construct all the sub-triangulations, which can be merged directly according to the property of the decoupled domain decomposition mode finally. Meanwhile, the streaming computation is introduced so that the algorithm has a more excellent capability of space. The analysis and experiments show that the algorithm has a low memory footprint and is efficient with a time complexity close to O(nlg(δ))(δ is the number of points in a cutting block and δ ≤ n).
TIN DEM data of the key areas have significant protection value. According to the data organization form, this paper proposes a new method of TIN DEM information disguising based on permutation-substitution theory which is a classical theory in cryptography. Firstly, the original data is permuted using the chaotic sequence which is generated by Tent map in frequency domain. Secondly, in the basis of Chinese remainder theorem, the final disguised data could be obtained by the numerical substitution in spatial domain, and the method of information reduction is also discussed in this paper. Finally, the disguising effectiveness and security performance of the method are validated by analyzing the experiments, which can meet the requirements of TIN DEM information disguising and also could guarantee the storage and transmission security for the TIN DEM data.
In the emergency treatment, multi-source resource environmental data, with isolation and poor relation, is difficult to be integrated retrieved. From the practical problems of applications, this article brings multi-source resource environmental data into an integrated grid reference system. Based on grid code, multi-source resource environmental data is integrated organized logically and integrated retrieval would be realized. With shape as the distinction, grid codes are given, and storage in code index table. According to code computing method, multi-source resource environmental data could be integrated retrieved. Real experiments show that, multi-source resource environmental data integrated retrieval is realized, and improve the efficiency by about 10 times.
Concerning that current automatic delimitation of land border dispute areas neglects some important factors, this essay puts forward a new automatic generation method of land delimitation line based on parallel simulated annealing algorithm. Firstly, "point-point" topological relationship is built up based on terrain line network and disposed according to delimitation laws. Secondly, coding and estimation of delimitation line is needed for simulated annealing algorithm, as well as generation of initial delimitation line. Thirdly, parallel simulated annealing algorithm need to combine different ways of annealing to search for the best delimitation line fast and fully. The experiment shows that this method can not only take area ratio decided by related countries, terrain and special areas into account, but also ensure that one could get the biggest resource profit, which can protect its delimitation profit in a better way.
In order to solve space scene similarity measure problem when the space scene contains differententity numbers, this article uses feature matrix to describe space scene, then it also uses the feature matrix of query scene and database scene to generate scene associated graph and uses the various matching circles of the associated graph to get space scene collection. After that, based on space scene completeness and similarity measure model for each scene of the scene collection, the matching degree will be calculated. Finally, we can get the best match scene and the matching results will be analyzed and evaluated.Experimental results show that this method can better measure the similarity space scene which contains differententity numbers.
We introduce a shape matching method based on the shape context, and apply it to the morphing of linear features. For the two representations of the same linear feature in two different scales, to get the correspondences of the point, the shape context of every point on the shape is calculated at the very beginning. Then we compute the similarity between different points using shape histogram. The two linear features are divided into two groups of sub-segments by the matching result. Finally, using the linear interpolation method to Morphing every pair of the corresponding sub-segments. The experimental results show that during the process of matching no special landmarks or key-points of the linear features are needed. It has strong adaptability to the shape matching and greatly improve the accuracy of morphing transformation.
The monitoring stations for air quality index (AQI) are sparsely distributed, and spatial interpolations are less accurate from the existing methods. A new algorithm is proposed based on the extended field intensity model. The single parameter model controls intensity attenuation by parameter c, while the optimal c value is computed from the relationship between c and deviation data with binary search method. The double parameters model adjusts intensity range by additional parameter k, while the optimal c and k are computed from the relationship among c, k and deviation data with iterative bilinear interpolation method. The 50 monitored sets of AQI value are taken as experimental data from Beijing, Tianjin, Wuhan and Zhengzhou Between August 2014 and April 2015. Based on cross validation and evaluation criteria RMSE, AME, PAEE, both single parameter model and double parameters model are implemented with their optimal parameters, then the extended field intensity model is compared with Kriging and inverse distance weighted methods. Experimental results prove that the precision of AQI interpolation from our algorithm is higher, while double parameters model obtains the highest precision. Our algorithm is suitable for spatial interpolation of sparse data with fixed number and locations, and can be used for spatial data with other types and dimensions.
We use the data of mobile across-fault deformation at main tectonic areas in China up to the end of 2014, aided by analysis on time-space evolution for anomalies of fault deformation and the index of feature intensity and the result of GPS velocity field, to search the variation background of tectonic activities in large area and short-term precursor anomalies, possibly being related to the activities of strong earthquakes such as Lushan, Minxian-Zhangxian, Ludian and Jinggu Ms 6.5~7.0 earthquakes during 2013~2014 at the eastern margin of Qinghai-Tibet plateau. The results show that, ① this group strong earthquakes during 2013~2014 possibly had relationship with the general accelerating of tectonic activities at the eastern margin of Qinghai-Tibet plateau. ② Lushan Ms 7.0 and Minxian-Zhangxin Ms 6.6 earthquakes were located inside this monitoring area, reflecting obvious short-term precursor anomalies; But Ludian Ms 6.5 and Jinggu Ms 6.6 earthquakes outside this monitoring area reflected weaker short-term precursor anomalies.
Rb clock products of BDS on-orbit satellites in 2013 are used to analyze the satellite clock qualities such as frequency stabilities and clock noise level. Quadratic polynomial model is used to fit BDS satellite clock offsets. The short-term frequency stabilities are achieved using Hadamard total variance method. Based on above indexes, BDS satellite clock characteristics which include phase, frequency, frequency shift and residual indexes have been analyzed. It can be concluded by the actual numerical examples that, the frequency stabilities of BDS satellite clocks in ten-thousand seconds is about 10-13. The stabilities of GEO satellite clocks are worse than that of the other satellite clocks. The satellite clocks of PRN04 and PRN08 have phase jumps during running time, and the stabilities of these satellite clocks have been enhanced after jump times. The stabilities of other satellite clocks are relatively steady. In general, the qualities of MEO satellite clocks are better than that of GEO and IGSO satellite clocks.
Multi-GNSS combination is helpful to improve the accuracy and relibality of satellite navigation and positioning. However, in case of BDS/GPS combined carrier phase positioning systems, it is difficult to fix all the ambiguities due to the increasing number of ambiguities, high measurement noises or residual atmosphere delays with the traditional Lambda method. But it is of greater probability to fix a subset of ambiguities. In this paper, we divided current partial ambiguity fixing methods into three categories and analyzed the characteristics of every method. Finally, the effect of the three partial resolution methods were tested with the measured BDS/GPS data. The results show that the success rate and ratio value is obviously improved when using partial ambiguity fixing, at the same time, the initialization time of RTK is shortened, the precision of kinematic positioning is also improved.
A new method of GPS satellite clock bias prediction based on exponential smoothing method (ESM) is presented in this paper. This new method can develop the prediction model successfully by using a small amount of data and has the advantages of easier calculation and convenience in operation. And the good results can still be acquired by this new method when the relevant historical data are absent or the changing trend of data is unobvious or unstable. By contrast with the quadratic polynomial model (QPM) and gray system model (GM) which are usually used in GPS satellite clock bias prediction, the calculating and analyzing results indicated that the ESM can be used in the medium-term and short-term prediction of GPS satellite clock bias and the prediction precision can reach up to nanosecond (ns) level. The prediction results of ESM are better than QPM but on the same level with GM when a small amount of data is used to establish the prediction model. And in the mean time, the ESM can also be used in the long-term prediction of GPS satellite clock bias and the prediction precision can reach up to microsecond (μs) level which is on the same level with GM.
Method of point-mass model is investigated thoroughly. The ill-posed problem in parameter estimation is solved by Tikhonov regularization. Water storage variation in China and adjacent is recovered by method of point-mass model using GRACE time-variable gravity field data. The inversion result is verified by GLDAS hydrological model data and method of spheric harmonic coefficient. Meanwhile the water storage variation series of 4 characteristic points are calculated. The results show that because of the separated influence of different region mass anomaly on gravity field, the signal of water storage variation in local area which is recovered by method point-mass model is more powerful. By comparing the water storage variation series of 4 characteristic points, it is concluded that the correlation between the recovery results by method of point-mass model and GLDAS hydrological model data is stronger.
The response of the solid Earth's crust to mass loading on the Earth's surface usually can be computed by Green's Function or Spherical Harmonic Function methods. The two methods are equivalent in mathematics, but will give different crustal deformation results from the same surface loading. We thus use surface fluid mass change data to quantitatively analyze the precision of the two methods' results in calculating the crustal deformation on the Earth's surface. The results indicate that the crustal deformation precisions of both methods, in the level of observed GPS error, are consistent. The two methods also have the same calculation efficiency for single station crustal deformation. However, in calculating the crustal deformations of global 1°×1° grid sites, the efficiency of the spherical harmonic function method is nearly 100 times faster than that of Green's function method. For the surface fluid change data, only using 2°×2°spatial grid data will have enough accuracy to correct the crustal deformation of GPS stations. The vertical crustal deformation caused by surface fluid changes can explain more than 50% variance of that of GPS stations globally.
GNSS (global navigation satellite systems) time uses atomic clock as the time standard while GNSS receiver uses crystal oscillator. The stability of atomic clock is several times of magnitudes better than crystal oscillator. Therefore, the intersystem time offset between different GNSS is more stable than receiver clock bias. It will benefit the positioning results if we can fully use that prior information. The purpose of this contribution is to analyze how to use that prior information properly and examine the influence on it. We deduce two positioning model based on different clock bias estimation method and introduce two parameter estimation algorithm:least square (LS) and extended Kalman filter (EKF). Static and dynamic tests were carried out and some conclusions are made:the LS cannot improve the positioning performance with the prior information of that the intersystem time offset between different GNSS is more stable than receiver clock bias. Static test results show the EKF itself has some kind of noise suppression. The noise is less in case we use the "stable" prior information. Dynamic test shows divergence of EKF can be suppressed by using the "stable" prior information.
The terrestrial water storage changes (TWSC) in Tianshan Mountains, Xinjiang of China was inferred using the GRACE monthly gravity data from 2003 to 2013. The noise existed in GRACE data was filtered by the combination of de-correlated technique and 300km Gaussian filter, and the GRACE post-process errors was reduced by the scaling factor, and the effect caused by GIA was cut down by the model provided by Paulson. The final GRACE result is consistent with the GLDAS and CPC hydrological models. The preliminary result indicates that the water storage present a decreasing rate which is about -0.54±0.27 mm/a in the Tianshan region, However, the TWSC fluctuates largely, specially, it shows a significant decrease in October 2008, which coincides with the drought conditions in the same period. The spatial distribution of TWSC shows a downward trend in the most parts of Tianshan region, with the maximum rate of -5.6±1.6 mm/a appearing in the middle of the Tianshan Mountains.