The Journal of Geomatics and Information Science of Wuhan University as an open journal, relies on the advantages of discipline of engineering surveying and mapping of China. It's aiming to be one of excellent Chinese Surveying Journals. The Journal mainly publishes research achievements on Digital Photogrammetry, Remote Sensing, Cartology and Geographical Information Systems, Satellite Geodesy, Physical Geodesy and Geodynamics, Surveying Engineering and Image and Graphics etc.
Objectives: Unstable approach is one of the most important risks that threaten the flight safety during the descending phase, and is affected by factors such as meteorological and topographical that have strong spatial heterogeneity. Methods: In this article, we use the big data of quick access recorder (QAR) that collected by China Academy of Civil Aviation Science and Technology (CAST) to detect the unstable approach happening in Airbus and Boeing aircrafts in January 2018, and utilize exploratory spatial data analysis methods to explore the spatial patterns and relative influencing factors. In addition, Pearson correlation coefficient and geographically weighted correlation coefficient are used to investigate the pair-wise relationships between unstable approach and factors. Results: Experimental results show that unstable approaches of different aircraft types are distinctive in spatial distributions, and unstable approach of same air-craft type spatially varies in different regions in China. Besides, the correlation description in the fine-grained scale is better than in the global scale. Conclusions: This study provides an important research basis and guidance for the future quantitative cause analysis, which is significant for avoiding such risks.
Objectives: Micro landform classification of grid digital elevation model(DEM) is the foundation of digital landform refinement application and has broad application prospects in pedology, urban planning, civil engineering, military affairs, diplomacy, etc. However, problems, e.g., low degree of automation and incomplete classification, are still existed in the micro landform classification method of grid DEM based on regular knowledge. With the advantages of convolutional neural network(CNN), a CNN method for grid DEM micro landform classification is constructed and its automated implementation process and approach are created. Methods: Taking the advantage of CNN that can automatically learn the sample data and mine the hidden knowledge of the data set, it can automatically learn the features of the micro landform data set and realize the automatic classification of the micro landform combining with the micro landform data. Firstly, the plane curvature, section curvature, slope and elevation of grid DEM in the experimental area are extracted as data sets. According to the decision table and prior knowledge, the hill-position of the sample data set is divided into 6 types, including summit, shoulder, back-slope, foot-slope, toe-slope and alluvium. Among which, 2/3 data are used as the training set, and the remaining sample data are used as the test set. According to the data requirements of the CNN and the characteristics of the data set, a convolution layer is constructed, which can automatically learn the data characteristics in training set, determines the training model, and uses the test set to evaluate the model’s accuracy. After the model meets requirements of the accuracy, the new micro landform factor generalizes the model are entered and the results of landform classification are outputted, i.e., the model is used to automatically classify hill-position of the grid DEM data. Results: CNN is used for grid DEM micro landform classification. The experimental results show that the overall accuracy of the model test dataset established by the selected typical sample points in the experimental area reaches 97.95%, and the Kappa index is 0.974 9, which demonstrates the applicability of the CNN algorithm in micro landform classification. CNN method has a strong dependence on landform category knowledge contained in typical sample points. Incomplete description of local shape category feature knowledge will affect the CNN algorithm in mining implicit knowledge directly, resulting in inconsistent adaptability to 6 types of micro landform. The highest classification accuracy of alluvium is 99.64%, while the lowest is toe-slope with 92.95%, and the accuracy rates of summit, shoulder, back-slope and foot-slope are 95.89%, 99.58%, 99.08% and 98.87%, respectively. The incomplete degree based on superposition analysis is 35.91%, while the method based on CNN does not exist incomplete. The reason is that superposition analysis is a classification method of multi-factor system for each grid node. For different regions, the landform complexity is inconsistent, and the incomplete degree range is uncertain. The CNN method takes its advantage of back propagation to solve the problem of incomplete classification. The classification accuracy of CNN method for micro landform classification of grid DEM data with different pixel sizes is different, among which, the average classification accuracy of the pixel size of 5 m is the highest with 97.67%. The classification accuracy rate with the pixel size of 10 m is relatively lower with 96.09%. The grid DEM with the pixel size of 5 m has the lowest accuracy dispersion of classification results of each category among the 6 types of hill-position, and shows strong adaptability to all type of land-form. In summary, grid DEM with the pixel size of 5 m is more suitable for automatic classification of micro landform based on CNN. Conclusions: Automatic classification of hill- position realized by CNN model is suitable for grid DEM micro landform classification. Compared to the classification method based on rule knowledge, the complicated data overlay analysis process is avoided and the completeness is improved. The overall accuracy of the model reaches 97.95%, among which, the highest classification accuracy is alluvium with 99.64%, while the lowest is toe-slope with 92.95%. Considering the influence of the pixel size of the grid DEM data on the classification accuracy, the grid DEM of 5 m × 5 m is more adaptable than2.5 m × 2.5 m and 10 m × 10 m.
Objectives: The sewer pipeline network is an essential urban infrastructure. It is easy to be with faults because of the complicated circumstance under the ground, as well as undertaking the transportation and circulation of water. To avoid serious accidents caused by the faults, periodic and systematical inspection devices and methods are necessary. While current methods exist a variety of limitations, especially for large-scale and hard-inspected urban sewer pipe networks. Thus, we proposed a novel collaborative inspection method with high efficiency and availability, and low cost for urban sewer pipe networks. Methods: Firstly, we invent a novel device named pipe capsule equipped with video cameras, lights, and a data storage facility. It will be dropped into pipes, move along with water flow, and record the videos of the inner walls until salvaged by the staff from the downstream. Such that, the video is captured to inspect the underground pipe. Multiple capsules are cooperated to complete the inspection task. A collaborative inspection method is designed through well design practice plan by maximizing the inspection range and degree, as well as mini-mizing the traveling length and time of the workers. Simulated annealing with a simple neighborhood search strategy is used to find the best practice plan. Results: Taking the sewer pipe network at Shenzhen University, China an example, an experiment was conducted to verify the performance of the proposed approach. The results demonstrate that the collaborative inspection method achieves the same inspection with a shorter working time. Conclusion: This study presented a collaborative inspection approach to detect the large-scale and complicated urban sewer networks with machine vision. It will benefit the inspection and operation of the urban underground pipe network.
Objectives: GNSS tomography technique is one of the most important methods to obtain the three-dimensional water vapor information. However, due to lack of enough initial prior information in the process of building the tomography model, the design matrix of tomography model is unstable and the tomographic result is poor, which has become an urgent problem to be resolved. Methods: First of all, the grid data derived from European Center for Medium-Range Weather Forecasting (ECMWF) is used to calculate the initial value of water vapor density in every voxel of interest area. And then, sophisticating the traditional tomography modeling using the calculated initial water vapor values. Finally, the influence of weightings of different equations in tomography model on tomographic result is also considered. Results: Comparing to the traditional methods, the proposed approach can enhance the accuracy of tomographic result by 41.2%, and its root mean squared error (RMSE) decreases from 1.82 g/m
3 to 1.07 g/m
3. Additionally, the mean absolute error (MAE), Bias and standard deviation (STD) also show a better performance than those of the traditional methods. Conclusions: The purpose is to improve the accuracy of tomographic water vapor profiles and used to sophisticate the established tomography model.
Objectives: Marine gravimetry is a critical method for obtaining high-precision, high-resolution earth gravity field information in the marine areas. For geoid improvement, the absolute reference of the gravity is of particular interest (in contrast to, e.g., exploration geophysics, where relative accuracy, or pre‑cision, is sufficient). Therefore, a constant drift behavior of the gravimeter and also the reference readings in the harbor are important for this application. Methods: The processing strategy of marine gravimetry in the Baltic Sea is studied. The drift function based on gravity differences at crossover points is used to reduce the influence of the abnormal gravimeter drift and the measurement accuracy reached 0. 5 mGal. In 2017, the drift stability of the air-marine gravimeter Chekan-AM is significantly improved after an update. Besides, the real measured gravity gradients instead of the normal gravity gradient (0.308 6 mGal/m) in the harbors are used to calculate the gravity differences between the Chekan-AM sensor and connection points due to their large altitude differences in the campaign URD2017. Results: The gravimetry data processing results show that the accuracy of the marine gravimetry onboard the conventional research vessel is 0.2 mGal and that onboard the ferry is 0.6 mGal. The accuracy reaches the aim of sub-mGal in the project of Finalising Surveys for the Baltic Motorways of the Sea (FAMOS) and is even close to the current highest accuracy level of 0.1–0.2 mGal in marine gravimetry. The preliminary results of the construction of the local geoid in the Baltic Sea region are given, which proves that the obtained marine gravity measurement data plays an important role in filling data gaps, verifying old gravity data and improving the accuracy of the regional geoid. Conclusions: Our research provides the foundation of marine gravimetry technologies and gravity data sets for establishing a unified geoid for the entire Baltic region in the future.
Objectives: The load balancing of map information under multiple constraints is one of the difficulties in cartographic generalization. In small and medium-scale maps, it is necessary to comprehensively consider their administrative levels, topologies and metric relationships for the scale conversion of townships and village residential point features to make the map information load reasonable at a certain scale. Methods: This paper proposes a method for selecting residential point features based on optimal information entropy constraints. Under the constraints of optimal information entropy, the metric relationship constraints are adjusted, the semantic relationships are prioritized, the residential point features with higher administrative levels are reserved, and for the residential point features with lower administrative levels, if they are not the endpoints of the road and do not satisfy the metric relationship constraint, then the points are deleted, and the process is iterated until the optimal information entropy constraint is satisfied. Results: Experiments with 1:250 000 residential point data have realized the selection of residential location elements that maintain topological consistency, level priority, and metric rationality. The load balancing and readability of the map are effectively maintained, meanwhile, the amount of effective information of the map is maximized based on the algorithm. Conclusions: The optimal information entropy constraint is adopted for the selection of residential points, which can retain the density characteristics of the spatial distribution of the residential point group as a whole, and achieve the load balancing of map information in effect.
Objectives: Face age estimation is still challenging because facial age image are not only affected intrinsic factors (e.g., genes), but also external factors (e.g., living conditions). Therefore, it is difficult to find features which can describe the variation of age accurately. Methods: We propose a robust face age analysis method based on multi-layer sparse representation, which combines the respective characteristics of active appearance model (AAM), local binary patterns (LBP) and bio‑inspired features (BIF). This method will fully consider the way of object recognition of the human, the similarity of adjacent ages and the classification principle of signal sparse representation. In addition, in order to reduce the interference from the face identity factor, the two‑factors‑analysis method is proposed to separate the face identity factor. Results: A large number of experiments are conducted on the public FG‑NET, IFDB and MORPH2 datasets, and the mean absolute errors are within 4.65 years and 3.64 years for FG‑NET and MORPH2 database, respectively. Conclusions: The experimental results show that the multi-layer sparse representation model has strong discrimination and robustness, which overcomes the shortcomings of traditional methods.
Objectives: With the acceleration of urbanization and the rapid expansion of urban population, the population data on the street scale plays an important role in urban economic, social, resource and environmental development. Therefore, estimating population on urban street scales with high-resolution remote sensing images is of great theoretical and practical value for promoting sustainable urban development. Methods: This paper uses the building information in remote sensing images and census data to analyze the correlation between street building information and population data, and the city street scale population estimation models based on building information are proposed. The best estimation model for the number and geometric characteristics of buildings and population is established based on the multivariate stepwise regression method and the Akaike information criterion that determine the significant feature variables of the building. Results: The Experiments show that the population estimation model proposed in this paper can estimate the population of street scale with high accuracy. On the one hand, the floor data can be directly applied to the estimation model. On the other hand, the floor data is also the basis for classifying street buildings. Dividing street buildings into low-rise, middle-rise and high-rise buildings according to the number of floors can significantly improve the population estimation effect of the model. Conclusions: The population estimation model proposed in this paper can estimate the population of street scale with high accuracy. This paper uses street-scale data in administrative divisions, but the estimation range of the proposed model is theoretically also valid in population estimation across street areas, which needs to be verified in subsequent studies.
Objectives: Pyramid scene parsing net (PSP Net), a kind of neural network with deep structure, extracts the features of remote sensing image better than the traditional model, e.g., artificial neural network, and supports vector machine. This algorithm can embed the contextual features of difficult scenes into the pixel prediction framework of the full convolutional neural network (FCN), fully understand the scene, realize accurate prediction of each pixel category, location and shape, and fuse local and global information together to propose an optimization strategy for moderate supervised loss. A deep learning algorithm based on PSP Net is proposed to achieve higher accuracy in image classification and effectively promote remote sensing image automation and intelligent interpretation. Methods: Conventional architecture for fast feature embedding (CAFFE) is a deep learning framework with is expressive, fast and thought modular, and supports a variety of deep learning architectures for image classification and segmentation. PSP Net model under the CAFFE framework is used by modifying and optimizing the network to eliminate the over-fitting effect. Using the remote sensing image of Hubei Province as the experimental data, the land with the30 m resolution is studied with the aid of the analysis ability of the context scene of PSP Net. The Python program is used in the CAFFE depth learning framework for operation. The operation network obtains the global feature information through the core structure (pyramid pooling module) and completes the generation of the data set. In the experiment, the 507 standard partial images of 900 × 600 pixels in the landsat image of Hubei Province are used, and the sample sets suitable for depth learning are generated by preprocessing. Among these, 300 data are selected, including 240 training sets, 44 prediction sets, and 16 validation sets. The overall accuracy (OA) index is used as the prediction accuracy index of the preliminary evaluation model to evaluate the accuracy of the prediction model. Results:PSP Net model under the deep learning framework of CAFFE is used to train the sample data. The learning rate of 10 × 10
−10 is set, and the million times training model is selected to protect against overfitting of the data. Through the generaliza‑tion of the model and the generalization and iteration of the samples, the land cover of the 3 phases of TM im‑ages about Hubei Province are classified, and the classification accuracy is 82.2%, 83.4%and 83.7%, respectively. Conclusions: The results show that the land cover classification of remote sensing images can be realized quickly, effectively and accurately by PSP Net.
