Table of Content

20 June 2019, Volume 48 Issue 6

Previous Issue    Next Issue

Photogrammetry and Remote Sensing

Improvement strategy for location accuracy without ground control points of 3rd satellite of TH-1

WANG Renxiang, WANG Jianrong, LI Jing, ZHU Leiming, LI Wu, YANG Junfeng

2019, 48(6):  671-675.  doi:10.11947/j.AGCS.2019.20190058

Asbtract ( )   HTML   PDF (850KB) ( )  

References | Related Articles | Metrics

The transmission photogrammetric satellite with global coverage on a continuous basis belongs to dynamic photography, and the location without ground control points (GCPs) can-not reach to the level of the frame imagery. In this paper, the location without GCPs is researched in-depth about 3th satellite, based on the existing research results of TH-1. In order to counteract the negative effect of small width-to-height ratio (ratio of satellite strip photographic coverage width to satellite orbital height), the additional corrections of interior orientation elements were performed during the on-orbit calibration. As a result, the systematic errors were reduced while the horizontal location accuracy was improved, and the location accuracy without GCPs have achieved to 3.7 m (horizontal accuracy) and 2.4 m (vertical elevation accuracy) using foreign testing field. The location without GCPs of a single strip satellite imagery with global coverage on a continuous basis can be improved to the theoretical level by using its GNSS receiver, star tracker and domestic calibration results of camera parameters.

Using spatial-spectral regularized hypergraph embedding for hyperspectral image classification

HUANG Hong, CHEN Meili, WANG Lihua, LI Zhengying

2019, 48(6):  676-687.  doi:10.11947/j.AGCS.2019.20180469

Asbtract ( )   HTML   PDF (11684KB) ( )  

References | Related Articles | Metrics

In recent years, many graph embedding methods were developed for dimensionality reduction (DR) of hyperspectral image (HSI), while these methods only use spectral information to reveal a simple intrinsic relation and ignore complex spatial-spectral structure in HSI. A new DR method termed spatial-spectral regularized sparse hypergraph embedding (SSRSHE) is proposed for the HSI classification. SSRSHE explores sparse coefficients to adaptively select neighbors for constructing the regularized sparse intrinsic hypergraph and the regularized sparse penalty hypergraph. Based on the spatial consistency property of HSI, a local spatial neighborhood scatter is computed to preserve local structure, and a total scatter is computed for global structure of HSI. Then, the optimal discriminant projection is obtained by possessing better intrinsic data compactness and penalty pixels separability, which is beneficial for classification. The experimental results on Indian Pines and PaviaU hyperspectral data sets show that the overall classification accuracies respectively reach 86.7% and 92.2%. The proposed SSRSHE method can effectively improve classification performance compared with the traditional spectral DR algorithms.

Single and multiple rotation averaging iterative optimization coupled 3D reconstruction for low-altitude images using SfM algorithm

HE Haiqing, CHEN Min, CHEN Ting, LI Dajun, CHEN Xiaoyong

2019, 48(6):  688-697.  doi:10.11947/j.AGCS.2019.20180063

Asbtract ( )   HTML   PDF (3223KB) ( )  

References | Related Articles | Metrics

To reduce the cumulative errors caused by initial stereo model construction and related images augmentation in the existing structure from motion based methods, a single & multiple rotation averaging iterative optimization coupled 3D reconstruction method for low-altitude images is proposed. Firstly, an initial undirected graph of correlated images is established based on images relative orientation. Strongly correlated images are found in terms of the optimal incremental decision function, which is constructed based on multiple factors. Secondly, single rotation averaging considering gross erroris used to expand images directed graph in local reference framework, the global uniform optimal solution of rotation and translation matrices can be computed by quaternion supported multiple rotation averaging without involving 3D points. Finally, bundle adjustment is used to refine 3D points, rotation and translation matrices.The experimental results show that the proposed method can achieve more accurate exterior orientation elements and recover more 3D object points, and significantly improve the efficiency of the incremental structure from motion.

Deep metric learning method for high resolution remote sensing image scene classification

YE Lihua, WANG Lei, ZHANG Wenwen, LI Yonggang, WANG Zengkai

2019, 48(6):  698-707.  doi:10.11947/j.AGCS.2019.20180434

Asbtract ( )   HTML   PDF (2435KB) ( )  

References | Related Articles | Metrics

Due to the similarity of intra-class and dissimilarity of inter-class of high-resolution remote sensing image scene, it is difficult to identify some image scene class. In this paper, a new classification approach for high-resolution remote sensing image scene is proposed based on deep learning and metric learning. Firstly, a clustering center of each class is preset on the output features of deep learning model. Secondly, the Euclidean distance method is used to calculate the average central metric loss. Finally, the final loss function consists of a central metric loss term, a cross entropy loss term, and a weight and bias term. The goal of this method is to improve the classification accuracy by forcing intra-class compactness and inter-class separability. The experimental results show that the proposed method significantly improves the classification accuracy. Compared with state-of-the-art results, the classification accuracy ratios on RSSCN7, UC Merced and NWPU-RESISC45 datasets are increased by 1.46%, 1.09% and 2.51%, respectively.

3D reconstruction with inverse depth filter of feature-based visual SLAM

ZHANG Yi, JIANG Ting, JIANG Gangwu, YU Anzhu, YU Ying

2019, 48(6):  708-717.  doi:10.11947/j.AGCS.2019.20180421

Asbtract ( )   HTML   PDF (15916KB) ( )  

References | Related Articles | Metrics

Aiming at the problem that the current feature-based visual SLAM can only reconstruct a sparse point cloud and the ordinary frame does not contribute to point depth estimation, a novel 3D reconstruction method with inverse depth filter of feature-based visual SLAM is proposed, which utilizes video sequence to incrementally build a denser scene structure in real-time. Specifically, a motion model based keyframe tracking approach is designed to provide accurate relative pose relationship. The map point is no longer calculated directly by two-frame-triangulation, instead it is accumulated and updated by information of several frames with an inverse depth filter based on probability distribution. A back-end hybrid optimization framework composed of feature and direct method is introduced, as well as an adjustment constraint based point screening strategy, which can precisely and efficiently solve camera pose and structure. The experimental results demonstrate the superiority of proposed method on computational speed and pose estimation accuracy compared with existing methods. Meanwhile, it is shown that our method can reconstruct a denser globally consistent point cloud map.

A remote sensing image semantic segmentation method by combining deformable convolution with conditional random fields

ZUO Zongcheng, ZHANG Wen, ZHANG Dongying

2019, 48(6):  718-726.  doi:10.11947/j.AGCS.2019.20170740

Asbtract ( )   HTML   PDF (8857KB) ( )  

References | Related Articles | Metrics

Currently, deep convolutional neural networks have made great progress in the field of semantic segmentation. Because of the fixed convolution kernel geometry, standard convolution neural networks have been limited the ability to simulate geometric transformations. Therefore, a deformable convolution is introduced to enhance the adaptability of convolutional networks to spatial transformation. Considering that the deep convolutional neural networks cannot adequately segment the local objects at the output layer due to using the pooling layers in neural networks architecture. To overcome this shortcoming, the rough prediction segmentation results of the neural network output layer will be processed by fully connected conditional random fields to improve the ability of image segmentation. The proposed method can easily be trained by end-to-end using standard backpropagation algorithms. Finally, the proposed method is tested on the ISPRS dataset. The results show that the proposed method can effectively overcome the influence of the complex structure of the segmentation object and obtain state-of-the-art accuracy on the ISPRS Vaihingen 2D semantic labeling dataset.

A template matching method of multimodal remote sensing images based on deep convolutional feature representation

NAN Ke, QI Hua, YE Yuanxin

2019, 48(6):  727-736.  doi:10.11947/j.AGCS.2019.20180432

Asbtract ( )   HTML   PDF (12671KB) ( )  

References | Related Articles | Metrics

Due to significant non-linear radiometric differences between multimodal remote sensing images (e.g., optical, infrared, and SAR), traditional methods cannot efficiently extract common features between such images, and are vulnerable for image matching. To address that, the deep learning technique is introduced into the present study to design a matching method based on Siamese network, which aims to extract common features between multimodal images. The network is first optimized by removing the pooling layer and extracting the feature layer from Siamese network to maintain the integrity and positional accuracy of the feature information, making it possible the effective extraction of common features between multimodal images. Then, the template matching strategy is adopted to achieve high-precision matching of multimodal images. The proposed method is evaluated by using multiple multimodal remote sensing images. The results show that the proposed method outperforms traditional template-matching methods in both the matching correct ratio and matching accuracy.

Interferometric calibration method for spaceborne SAR based on independent parameter decomposition

FAN Jun, LI Tao, ZUO Xiaoqing, CHEN Qianfu, ZHANG Xiang, LU Jing

2019, 48(6):  737-746.  doi:10.11947/j.AGCS.2019.20180128

Asbtract ( )   HTML   PDF (5040KB) ( )  

References | Related Articles | Metrics

The accuracy of interferometric parameters of InSAR plays a significant role in precise digital elevation model (DEM). However, interferometric parameters are traditionally calculated jointly, making them difficult to decompose precisely. It is proposed that an interferometric calibration method based on independent parameter decomposition(IPD) by the authors.Firstly, the parameters related to the interferometric SAR measurement are determined based on the three-dimensional reconstruction model. Secondly, the sensitivity of interferometric parameters is quantitatively analyzed after the geometric parameters are completely decomposed. Finally, each interferometric parameter is calculated based on IPD and interferometric calibration model is established. Weinan of Shanxi Province is taken as an example area. Four TerraSAR-X/TanDEM-X image pairs are used to carry out the interferometric calibration experiment. The results show that the elevation accuracy of all interferometric results are less than 2.54 m for this experimental data after interferometric calibration. Furthermore, the proposed method can obtain the accuracy of DEM products less than 1.21 m in the flat area and 3.11 m in the mountain area respectively.That results can prove the correctiveness and effectiveness of the proposed IPD based interferometric calibration method. The results could provide a technical basis for topographic mapping of 1:25 000 scale in the flat area and mountain area.

Cartography and Geoinformation

Rapid mapping of emergency scenario and cartographic information transmission

DU Ping, LIU Tao, LI Dingkai, YANG Xiaoxia

2019, 48(6):  747-755.  doi:10.11947/j.AGCS.2019.20180414

Asbtract ( )   HTML   PDF (999KB) ( )  

References | Related Articles | Metrics

At present, the emergency response mode has been changed from "prediction-response" to "scenario-response". Emergency thematic map as an important content of emergency surveying and mapping, its production way should be changed according to the new mode. This paper introduces the definition of unconventional emergency scenarios and geographic scenarios, points out that geographic scenarios as the physical space in which various scenario elements exist or occur are the object of emergency mapping, discusses the principle and 4 main characteristics of rapid mapping of emergency scenario, proposes a cartographic information transmission model combining the traditional cartographic information transmission theory with the new emergency response mode, gives detailed explanations to the process of spatial information transmission and analyzes the spatial cognition of map senders and receivers which affects the efficiency of cartographic information transmission.

A visualization method of continuous area cartogram for two or multiple variables

LI Xiang, WANG Lina, ZHANG Weidong, YANG Fei, YANG Zhenkai

2019, 48(6):  756-766.  doi:10.11947/j.AGCS.2019.20180353

Asbtract ( )   HTML   PDF (16125KB) ( )  

References | Related Articles | Metrics

Area cartogram is a visualization method that quantitatively represents regional attribute information by using the area size. Area cartogram is more conductive to the bivariate/multivariate mapping because the area size itself participates in the expression of variables. Now, bivariate/multivariate mapping based on area cartogram is difficult to express the basic situation between adjacent regions, and it is also difficult to express the spatial distribution of different geographical phenomena, to detect differences between two or more variables and spatial patterns. A method of a continuous Area cartogram for two or multiple variables has been proposed in this paper. Firstly, compensation of grid density and the progressive heuristics of the integration step are used to improve and optimize the classic algorithm of continuous area cartogram-"the diffusion-based method for producing density equalizing maps". Then, the first variable is visualized by area cartogram and the second or more variables are visualized by interpolating location on a continuous Area cartogram and symbolization. Finally, we use the population density and bank/ATM distribution data in Munich (bivariate mapping), the population density, kindergarten distribution and scale data in Augsburg (multivariate mapping) as case studies. This method is proved to be more effective and superior by the experiment results.

A topological relations-based inconsistency detection method for spatiotemporal land cover objects in raster space

KANG Shun, CHEN Jun, PENG Shu

2019, 48(6):  767-779.  doi:10.11947/j.AGCS.2019.20180346

Asbtract ( )   HTML   PDF (2310KB) ( )  

References | Related Articles | Metrics

Land cover updates are key datasets in geographic national conditions monitoring, environmental change assessment, and ecological system protection. Remote sensing technology has become an important tool for an update of land cover datasets. However, as the complexity of spectrum, texture and temporal characteristics, omission and commission errors usually occur in land cover data, leading to the issue of spatiotemporal land cover object inconsistency. Currently, detection of land cover data inconsistency gives priory to manual inspection, and partial automation is implemented. In practice, a huge of workmen and time are needed in data inconsistency detection, lacking of an automated detection tool. The challenge of land cover data inconsistency detection in raster space is analyzed in this study. A logic quantifiers-based topological relationships calculation in raster space, the initial-judgment rules construction based on confidence interval, and a post-judgement using spatial multi-matching are proposed, forming a "relation-rule-judgment" detection system. Inconsistency detection of the GlobeLand30 datasets is conducted in study areas of Linqu and Kenli, Shandong, China. Comparing with statistics inconsistency detection, and taking remote sensing images as reference, the effectiveness of inconsistent land cover objects detected in update is validated, and this method is proved to be practically feasible.

The valley extraction based on the hexagonal grid-based DEM

WANG Lu, AI Tinghua

2019, 48(6):  780-790.  doi:10.11947/j.AGCS.2019.20180408

Asbtract ( )   HTML   PDF (11158KB) ( )  

References | Related Articles | Metrics

Valley extraction from grid-based DEM depends on the geometrical shape, spatial tessellation and placement of grids. Traditional researches ignore that the shape of grid can influence the neighborhood relationship, impacting on the result of valley extraction. The traditional D8 algorithm is based on eight-neighborhood relationship, where the distance difference between node neighborhoods and edge neighborhoods results in the imbalance of topographic change expression in these two directions and further affects the extraction results of the valley lines. Hexagonal grid has been proved to be advantageous over square grid due to its consistent connectivity, isotropy of local neighborhoods, higher symmetry, more compact, and so on. Taking these merits into consideration, this paper tries to explore the advantage of hexagonal grid-based DEM on the results of valley extraction in comparison with traditional square structure. First, preprocess for valley extraction with depression filling, flow direction calculation and the flow direction in flat area based on the six-neighborhood relationship. Then, connect the flow lines according to the estimated flow direction. From the comparison between the results of hexagonal grid-based DEM and square grid-based DEM, we conclude that the hexagonal grid-based DEM has a superior capability in maintaining the detailed shape and the characteristics of extracted valley networks. That is to say, with the decrease of resolution, the valley lines extracted by hexagonal DEM have better spatial agreement with real river systems, and the shape loss characteristic of them is lower. In conclusion, the hexagonal grid-based DEM has higher data accuracy, and it can extract a finer valley networks with same data volume.

Complex radix number modeling and encoding operation for the planar aperture 4 hexagon grid system

DU Lingyu, MA Qiuhe, BEN Jin, WANG Rui

2019, 48(6):  791-800.  doi:10.11947/j.AGCS.2019.20180372

Asbtract ( )   HTML   PDF (5274KB) ( )  

References | Related Articles | Metrics

Grid system is a multi-resolution raster data structure, which is widely applied in organization, processing and analysis of multi-scale geospatial data. Research on hexagon grid system with important geometric attributes has attracted extensive attention in academia. Description and calculation of hierarchical relation is one of the research difficulties. According to the complex radix number theory and the affiliation of grid cells in interval hierarchy, the mathematical model of the planar aperture 4 hexagon grid system is established. Based on these, the equivalent encoding scheme is proposed, the encoding operations are defined and the rules of them are generalized. Meanwhile, the coding index and transformation between code and Cartesian coordinates are designed. The results of contrast experiments show that the proposed encoding scheme has structural symmetry compared with similar schemes, which can significantly improve the efficiency of encoding operation and has practical application potential.

Summary of PhD Thesis

Research on registration and rendering method of video to enhance 3D scene

ZHOU Fan

2019, 48(6):  801-801.  doi:10.11947/j.AGCS.2019.20180315

Asbtract ( )   HTML   PDF (655KB) ( )  

Related Articles | Metrics

Oblique photogrammetry based scene 3D reconstruction with structure sensing functions

XIAO Xiongwu

2019, 48(6):  802-802.  doi:10.11947/j.AGCS.2019.20180319

Asbtract ( )   HTML   PDF (663KB) ( )  

Related Articles | Metrics

Study on the extraction of aerosol information and its spatial-temporal changes based on PARASOL and CALIPSOL remote sensing data in the Yangtze River Delta

CHENG Feng

2019, 48(6):  803-803.  doi:10.11947/j.AGCS.2019.20180328

Asbtract ( )   HTML   PDF (656KB) ( )  

Related Articles | Metrics

Visual-aided indoor mapping and pedestrian navigation research

LIU Tao

2019, 48(6):  804-804.  doi:10.11947/j.AGCS.2019.20180339

Asbtract ( )   HTML   PDF (654KB) ( )  

Related Articles | Metrics