Acta Geodaetica et Cartographica Sinica

A general model for compensating remainder dynamic environment effect on marine and airborne gravimetry

HUANG Motao, CHEN Xin, DENG Kailiang, OUYANG Yongzhong, LU Xiuping, WU Taiqi, ZHAI Guojun

2020, 49(2): 135-146. doi:10.11947/j.AGCS.2020.20190010

Asbtract ( )

HTML ( )

PDF (3850KB) ( )

References | Related Articles | Metrics

In view of the fact that there always existing remainder dynamic environment effect on marine and airborne gravimetry, after finishing a detail analysis on the mechanism of creating different kinds of measurement error source and their change characteristic, we propose a general model to compensate the remainder dynamic environment effect, which is suitable for different kinds of marine and airborne gravimeters. Optimum seeking method of general model expression and estimation method of model parameters are studied and discussed. The famous Akaike information criterion based on information theory is suggested to be used in optimum seeking method of general model expression. And a crosscorrelation analysis is used to make the estimation of model parameters. Finally, an optimum general model expression for compensating the remainder dynamic environment effect is determined under previous double restraining condition. An actual marine gravity measurement set is used as a case study to test the validity of the suggested method and model. It is showed that a distinct overall improvement is gained with the approach, the RMS error of crossover discrepancies is reduced from±9.35×10^-5m/s² with the original observed gravity data to±1.01×10^-5m/s² with the corrected data. This result embodies the excellent performance of the new method and model for eliminating high dynamic environment effect on marine and airborne gravimetry.

Predicting bathymetry by applying multiple regression analysis in the Southwest Indian Ocean Region

FAN Diao, LI Shanshan, YANG Junjun, MENG Shuyu, XING Zhibin, ZHANG Chi, FENG Jinkai

2020, 49(2): 147-161. doi:10.11947/j.AGCS.2020.20180526

Asbtract ( )

HTML ( )

PDF (8550KB) ( )

References | Related Articles | Metrics

Considering the fact that the sea floor topography and gravity anomaly or vertical gravity gradient anomaly show strong linear correlation in the corresponding frequency bands, the method based on using multivariate regression analysis technique to combine multi-gravity data to construct the seafloor model was proposed. Then, the inversion test and analysis were carried out in the part of SWIR(Southwest India Ridge) in the Southwest Indian Ocean. The results showed that the bathymetry model (BDVG model) based on multiple regression analysis has the highest accuracy compared with other models, which is 11.51% and 57.81% higher than the S&S V18.1 model and ETOPO1 model respectively. The accuracy of each bathymetry model is higher, and the relative error fluctuation is small where the water depth is above 2000 m, reflecting the good inversion effect in the deep sea area. In places where the seafloor is fluctuated drastically or in shallow seaarea, BDVG model has less variation in relative error and relative error fluctuation than BDG model and BVGG model established by gravity anomaly and vertical gravity gradient anomaly as a single input source, reflecting the BDVG model has better stability and the necessity and advantage of joint inversion. The only shaft-deficient rift oceanic ridge section (27 oceanic ridge section) on the Indomed FZ-Gallieni FZ is currently in the stage of sufficient magma supply, and the seafloor expansion has less influence on it. At the same time, due to the influence of the symmetric splitting, several rises are symmetrically distributed along the north and south of the axis.

Segmentation of polluted SSS image by combining NSCT and multifractal

HE Yicai, ZHAO Jianhu, ZHANG Hongmei, RUAN Shilun

2020, 49(2): 162-170. doi:10.11947/j.AGCS.2020.20180276

Asbtract ( )

HTML ( )

PDF (4533KB) ( )

References | Related Articles | Metrics

To improve the accuracy of current segmentation algorithms for the side scan sonar image with high noise, a side scan sonar image segmentation method is proposed that comprehensively utilizing of decomposing image by NSCT (non-subsampled contourlet transform), enhancing image by combination of local standard deviation and mean, estimating the singularity of image by multi-fractal. Firstly, NSCT is used to decompose images to obtain low frequency image which filtered out high frequency noise and retain contour information and a series of high-frequency direction sub-band images. Then, based on the feature that target shadow appeared with the target in side scan sonar images, it is calculated that the low-frequency image feature combined the local standard deviation and mean to obtain the feature images that highlight the characteristics of the target and its shadow respectively, use the multifractal method to segment the feature image to get the result of low-frequency image segmentation. The image difference and non-maximal suppression methods are used to segment the high-frequency direction sub-band images and obtain the high-frequency segmentation results. Finally, it is obtained that the fine edge of the target and its shadow by combing of high and low frequency segmentation result. The validity of this method is verified by experiments.

On-orbit geometric calibration and accuracy verification of GF-6 WFV camera

WANG Mi, GUO Beibei, LONG Xiaoxiang, XUE Lin, CHENG Yufeng, JIN Shuying, ZHOU Xiao

2020, 49(2): 171-180. doi:10.11947/j.AGCS.2020.20190265

Asbtract ( )

HTML ( )

PDF (8362KB) ( )

References | Related Articles | Metrics

The wide field of view(WFV) camera on GF-6 satellite can achieve the swath width up to 800 km by single camera, which has unique advantages for large-scale surface observation and environmental monitoring. On-orbit geometric calibration is a key technology of optical remote sensing satellite geometric processing, and directly affects the geometric quality of images. Considering the distortion characteristics of the ultra-large field of view and the imaging characteristics of multi-spectral bands for WFV camera, this paper proposes a method of on-orbit geometric calibration for WFV camera. The system errors of WFV camera are compensated by a geometric calibration model based on detector direction angle, and the calibration parameters of each band are estimated jointly by absolute calibration and relative calibration methods. Using Landsat 8 digital orthophoto image, ZY-3 digital surface modeland ASTER GDEM as reference data, the absolute calibration and relative calibration of WFV camera are processed. The geometric calibration results show that the absolute positioning accuracy of WFV image is about 3 pixels, the internal geometric accuracy is within 1 pixel, and the band-to-band registration accuracy is better than 0.3 pixels, which show that the geometric quality of GF-6 WFV camera has been significantly improved after on-orbit geometric calibration.

A GPU-PatchMatch multi-view dense matching algorithm based on parallel propagation

DENG Fei, YAN Qingsong, XIAO Teng

2020, 49(2): 181-190. doi:10.11947/j.AGCS.2020.20180459

Asbtract ( )

HTML ( )

PDF (8360KB) ( )

References | Related Articles | Metrics

Aiming at the problem of low efficiency of multi-view dense matching, a GPU-PatchMatch multi-view dense matching algorithm is proposed. The algorithm uses GPU to improve the computational efficiency of PatchMatch. At the same time, it also makes full use of sparse scene information to initialize the depth information. In addition, in order to improve the propagation efficiency, it uses the pyramid red-blackboard to propagate the depth information in parallel. Finally, the experiments are carried out on the DTU, Strecha and Vaihigen datasets, and compared with the commonly used multi-view dense matching algorithms. The results show that our algorithm has a significant improvement in reconstruction efficiency, which is 7 times higher than the CPU algorithm (such as PMVS, MVE and OpenMVS), and 2.5 times higher than the GPU algorithm (such as Gipuma), which proves the effectiveness of the proposed method.

Flood events process detection and near realtime service based on sensor web

DU Wenying, CHEN Nengcheng, YUAN Sai

2020, 49(2): 191-201. doi:10.11947/j.AGCS.2020.20180378

Asbtract ( )

HTML ( )

PDF (9374KB) ( )

References | Related Articles | Metrics

Full life cycle flood detection and service (FD&S) is of great significance to ensuring people's lives and properties. Flood detection methods generally focuses on the flood section/average state, lack of the overall understanding of the flood process from the occurrence, through the development, and to the end of floods, and the detection and service of floods are passive and lagging.This paper made the flood process detection rule and improved the water level prediction model, which were employed as theoretical foundations, and combined them with sensor web to construct the process-based FD&S (PFD&S) method.Based on the PFD&S method,this paper developed the PFD&S prototype, which consists of the sensor layer, the data access layer, the flood detection layer, and the user interaction layer, and has the two operating modes of data publishing and flood subscription. The floods occurring in the Huanghan basin, Hubei, China in the summer of 2016 were selected as the case studies to test the feasibility and validation of the PFD&S method and prototype. The results demonstrated that the proposed PFD&S method and prototype could precisely determine the flood phases, provide the water level prediction, alert, and information statistics services according to the requirements of different flood phases, and the PFD&S method featured instantaneity and extensibility.

Airborne LiDAR point cloud classification based on deep residual network

ZHAO Chuan, GUO Haitao, LU Jun, YU Donghang, ZHANG Baoming

2020, 49(2): 202-213. doi:10.11947/j.AGCS.2020.20190004

Asbtract ( )

HTML ( )

PDF (8457KB) ( )

References | Related Articles | Metrics

Airborne LiDAR point cloud classification is one of the key steps for three-dimensional reconstruction of urban scenes. To leverage the existing high-performing deep learning network model in image field of image processing, improve classification accuracy and reduce training time and demand for training samples simultaneously, an airborne LiDAR point cloud classification method based on deep residual network is proposed in this paper. Firstly, high discriminative low-level features, i.e. normalized height, point cloud normal vector, intensity and normalized differential vegetation index, are extracted. Secondly, by setting different neighborhood sizes and perspectives, multi-scale and multi-view point cloud feature images are generated via using the proposed point cloud feature image generation strategy. Then, point cloud feature images are input into the pre-trained deep residual network to extract multi-scale and multi-view deep features. Finally, a neural network classifier is constructed and trained, point cloud classification results are obtained by utilizing the trained classifier and postprocessing. Benchmark datasets of ISPRS 3D Semantic Labeling Contest are used, the experimental results show that the proposed method can effectively distinguish 8 types ground objects such as buildings, ground and vehicles etc., and the overall accuracy of the classification result is 87.1%, which can provide reliable information for three-dimensional reconstruction of urban scenes.

Super-resolution reconstruction of “straring” satellite video motion scene considering motion estimation error

BU Lijing, ZHENG Xinjie, ZHANG Zhengpeng

2020, 49(2): 214-224. doi:10.11947/j.AGCS.2020.20180544

Asbtract ( )

HTML ( )

PDF (14838KB) ( )

References | Related Articles | Metrics

In the application of super-resolution reconstruction of staring satellite video image, due to inaccurate motion estimation, the reconstructed moving objects will appear "Grid" And "Tailing". in this paper, a motion scene super-resolution reconstruction method considering motion estimation error is proposed. The sensitivity of the regularized model expressed by L1 and L2 norms to noise is analyzed under the condition of motion estimation error.In this paper, robust estimation method is introduced to express the fidelity term of regularized model, so as to reduce the influence of motion estimation errors on the reconstructed results. On the basis of robust estimation model, tikhonov, TV and BTV regular terms are added respectively, and the super-resolution reconstruction effect of each model is analyzed. A bilateral filtering super-resolution reconstruction method based on robust estimation is proposed. The experiment chooses skybox and jilin-1 satellite video data. The results show that the method proposed in this paper is effective in reconstructing moving objects,and the details of the ground objects are clearer. In terms of normalized difference and gradient energy evaluation index, this method is superior to the classical method.

A semantic enhancement method for photorealistic mesh model based on local parameterization

WANG Libin, HU Han, ZHU Qing, DING Yulin, CHEN Min

2020, 49(2): 225-234. doi:10.11947/j.AGCS.2020.20180588

Asbtract ( )

HTML ( )

PDF (11177KB) ( )

References | Related Articles | Metrics

With the advances in structure-from-motion and multi-view stereo, state-of-the-art oblique photogrammetric solutions can obtain city-scale photorealistic mesh models automatically. However, the mesh models are lack of fine geometric structure and semantic free. Aiming at solving this issue, it is proposed that a semantic enhancement method for photorealistic mesh models based on local surface parametrization. The basic idea behind the proposed method is that, through the representation of surface tree, it is converted that the seamless fusion of semantic components and photogrammetric mesh models to a replacing operation in a local area. The two 3D models are parametrized to 2D space in the local region and seamless merged and replaced in the UV space by 2D constrained delaunay triangulation (CDT). The replaced semantic components are then reversed transform to 3D space, which finalize the semantic enhancement automatically. Experiments on oblique images in Shenzhen reveal that the proposed method can effectively realize the automatic seamless fusion of semantic components with an open boundary and photorealistic mesh models. Compared with the commercial software Maya, based on the method of insertion and fusion, the proposed method has practical value for improving modeling efficiency.

Forest resource classification based on random forest and object oriented method

WANG Meng, ZHANG Xinchang, WANG Jiayao, SUN Ying, JIAN Ge, PAN Cuihong

2020, 49(2): 235-244. doi:10.11947/j.AGCS.2020.20190272

Asbtract ( )

HTML ( )

PDF (7169KB) ( )

References | Related Articles | Metrics

Given that there are few studies on forest resource classification with the lack of relatively simple and effective methods, a forest resource classification method integrating object-oriented segmentation and random forest is proposed in this paper. Object-oriented segmentation technology could efficiently reduce the "salt and pepper effect", and random forest classification algorithm has the advantages of high accuracy, strong anti-noise ability and satisfying stability. Therefore, we built the optimum random forest classification model by adjusting the object-oriented segmentation parameters, constructing the optimal feature space and estimating the number of decision trees in random forests. Besides, the SVM algorithm is taken into comparison. The results show that the overall accuracy of the classification algorithm in this study is 83.34%with the Kappa coefficient reaching 0.789 2, which are significantly higher than that of SVM algorithm. It proves that object-oriented random forest classification can effectively improve the accuracy of forest resource classification.

Visual clarity of vector curve and its application in web map generalization

AN Xiaoya, CHENG Xiaoqiang

2020, 49(2): 245-255. doi:10.11947/j.AGCS.2020.20190280

Asbtract ( )

HTML ( )

PDF (2439KB) ( )

References | Related Articles | Metrics

Public participatory map making is prone to visual problems such as visual coalescence, overcrowding, and crowdedness, which are only solved by automatic map generalization. Since both the original map scale and the target map scale are sometimes difficult to quantify accurately in the map, it is no longer applicable that the conventional map generalization method is based on the "original-target map scale" to judge whether or not the map generalization is needed. After visualizing the vector data, it will produce visual coalescence, the more noticeable the coalescence is, the worse the map representation is, and the more comprehensive the generalization demand is. Based on this rule, this paper proposes a quantitative description of visual coalescence and judges whether or not map generalization is needed. First of all, from the perspective of human visual perception, we designed a quantitative indicator of visual coalescence of vector curves-visual clarity. Then, based on the "pyramid" scale space, the clarity of the curve expressed in multiple scales is calculated, and the change function of the clarity is fitted. The experiment applies this function to web map generalization decisions for VGI geographic data. Experimental results show that this method can accurately determine whether each vector curve needs to be generalized, and can effectively solve the visual problems brought by the heterogeneity of geographic scale. At the same time, the clarity change function expands the scale description of the curve from a static value to a continuous function, which is expected to better support multi-scale spatial data processing and web map generalization.

Complex linear pattern recognition for regular pond groups

LIU Chengyi, WU Fang, GONG Xianyong, XING Ruixing, LUO Denghan

2020, 49(2): 256-266. doi:10.11947/j.AGCS.2020.20190030

Asbtract ( )

HTML ( )

PDF (14308KB) ( )

References | Related Articles | Metrics

The existing polygon linear pattern recognition method is incapable facing to the broken area in the regular pond group. In order to solve the problem, a complex linear pattern recognition method for the regular pond group is proposed. Firstly, the spatial characteristics and cognitive characteristics of the regular pond group are analyzed, and the multi-level cognitive order, which is from major-and-minor relationship, parallel relationship, linear pattern to complex linear pattern, is proposed. Secondly, the identification methods of major-and-minor relationship, parallel relationship, and linear pattern are designed. Finally, the identification model of complex linear pattern pond group is constructed. Experiments show that the proposed method can eliminate the adverse effects of broken area on the linear pattern extraction of regular pond groups, and effectively improve the quality of complex linear pattern recognition.

Research on meteorological early-warning model of landslides in Wenchuan earthquake area based on machine learning

HUANG Lu

2020, 49(2): 267-267. doi:10.11947/j.AGCS.2020.20190061

Asbtract ( )

HTML ( )

PDF (704KB) ( )

Table of Content