Fuzhou Duan

Application of unmanned airship image system and processing techniques for identifying of fresh water wetlands at a community scale

Wetland ecosystems have complex habitat characteristics due to the wetland existing between lands and water bodies, however, satellite remote sensing images have normally mixed pixels because of their spectral characteristics of the riverine surface feature, and it is very difficult to extract the specific information of aquatic ecosystems. Hence satellite remote sensing images are hardly to match the demanding from scientists for their geoinformation researches in wetland investigation with a higher spatial accuracy. Unmanned airship low-altitude remote sensing system can resolve this technique problem. Recent advances in photo-geometry and relational technologies have facilitated the use of inexpensive digital cameras with the airborne remote sensing. And the ability of an airship would be enable it to obtain high-quality spatial images at a very high resolution by utilizing low-cost imaging instruments such as CCD cameras. Due to few landmarks existing within the wetland areas, we set artificially signs as the control points in the study area. In this study, we obtained the scanned high-resolution images (with the spatial accuracy of 0.13m in a grid) using the unmanned airship platform in the Honghe National Nature Reserve (HNNR) located in the Northeast portion of the Sanjiang Plain. Using the software — Unmanned Aerial Vehicle (UAV) CCD Image Processing System, we have mosaic automatically the aerial photographs, and we also have mosaic artificially the images used the software ENVI 4.3 based on corresponding image points between adjacent images. The study results indicated that the auto-processing method relatively was more efficient image processing however with a lower accuracy (with 4 pixels as the RMS error). It meets a demand of the large amount of data process and a relatively lower accuracy. The artificial process method relatively has a higher accuracy (less 2 pixels as the RMS error) however time-consuming, which may be applied to the image process with a high accuracy however less data. The mosaic images is available to identify wetlands vegetation at a community scale, such as Populus davidiana, Betula platyphylla and Carex pseudocuraica, Carex lasiocarpa, Calamagrostis angustifolia. With the results from this study experiment, we can conclude that the airship images can be used effectively to identify complex wetland vegetation information at a community scale, it has a great significance for applying to both scientific study and practical purpose, such as wetland plant mapping, environment planning and landscape design of wetlands, wetland restoration and ecological protection, etc.

Collapsed houses automatic identification based on texture changes of post-earthquake aerial remote sensing image

Collapsed houses in city are one of the most important destructions after earthquake. The area, amount and rate of collapsed houses are the most essential data sources to decide how to reconstruct in the disaster area after earthquake. It is often difficult to access high-resolution remote sensing satellite images in time only with satellite sensors in the area. For example, after the Wenchuan earthquake on May 12 2008, high-resolution satellite remote sensing images had not been acquired after a few days in which all actions have to be taken for disaster mitigation and relief. With various aviation flight platforms, the aerial remote sensing images are obtained in time at a relatively short period of time with better weather in the earthquake zones. So how to rapidly identify and detect the damage building is the most important problem. In this paper, through the texture changes of aerial remote sensing images, the method of housing collapsed automatically determination is expounded. The approach for detecting the damaged buildings after earthquake is different to visual identification of the post-event aerial images. The approach utilizes the texture difference between the buildings and the collapsed buildings. It is based on an idea that if a building collapsed, it will produce more complex texture. After we calculate the texture complexity parameter, the building may has collapsed if the texture complexity parameter is beyond threshold. After Wenchuan earthquake, the aerial remote sensing images are testified by this way in Beichuan, Wenchuan and Anxian. The rate of houses collapsed and the ratio of the edge of texture are strongly correlated. If the rate is beyond 0.15, the house will collapse over eighty percent, and if the rate is under 0.10, the percentage drops. Therefore, through the ratio of the edge of pixels and the number of regional architecture of the region pixels, to estimate the rate of the houses collapsed is feasible and effective.

Color space transformation and object oriented based information extraction of aerial images

Low-altitude aerial remote sensing platforms accessed reality multi-color images which had obvious characteristics and fitted for visual interpretation. These images were lacking of spectral information but rich in shape and texture information. But, the reality was that there was less study on the automatic extraction of ground information from aerial images. In this paper, UAV images were selected as test data. By combining the object oriented method and the multi-resolution segmentation, the paper selected some effective characteristics, constructed the rule sets and classify the image into water, shrub, farmland, road, and house. Then, the result was compared with which obtained by maximum likelihood classification method. The results showed that: With the object-oriented method, it could get higher accuracies and efficiencies for actual applications, the overall classification accuracies and Kappa coefficient are more than 85%.

Research on the management and expression of disaster data

Natural disasters that frequently occur may cause enormous loss of human lives and damage of properties. However, these losses can be reduced effectively by establishing disaster control and reduction system. In recent years, disaster control and reduction system based on 3D GIS are paid more attention. 3D disaster reduction system has the ability not only in managing and analyzing spatial data as well as the traditional 2D disaster reduction system, but also in processing mass data and truly expressing the 3D visualizing of geographical environment. However, there are still some technical difficulties in disaster data management and disaster information effective expression of 3D disaster reduction system. This paper discusses the method of disaster data management and disaster information expression in 3D disaster reduction system. Based on this, this paper designs a method to manage disaster data in integrated environment according to the multi-source, multi-scale and other features of disaster data. Besides that, it also describes the expression way of several variety disaster data. The experiment verifies that the proposed methods in this paper have achieved a satisfactory result.

A method of airborne remote sensing image quality assessment based on image matching

Since the fast development of UAV remote sensing provides us low-altitude high-resolution remote sensing data which is in great need in China, the flight quality assessment becomes one of the most important procedures of aerial photogrammetry. This paper introduces a flight quality assessment method based on image matching, which is a new meaningful attempt. The method uses RANSAC to remove coarse matched points, and calculates image rotation angle, track overlap, forward overlap, side overlap and strip deformation according to the matching parameters. Then it assesses the quality of the Unmanned Aerial Vehicle(UAV) flight from five aspects mentioned above. The proposed method involves some key technologies such as image processing, accuracy calibration, photogrammetry, data query. In the end, it establishes a complete quality assessment system which is of great efficient and high precision.

Village-level multidimensional poverty measurement in China: Where and how

Village is an important implementation unit of national poverty alleviation and development strategies of rural China, and identifying the poverty degree, poverty type and poverty contributing factors of each poverty-stricken village is the precondition and guarantee of taking targeted measures in poverty alleviation strategies of China. To respond it, we construct a village-level multidimensional poverty measuring model, and use indicator contribution degree indices and linear regression method to explore poverty factors, while adopting Least Square Error (LSE) model and spatial econometric analysis model to identify the villages’ poverty types and poverty difference. The case study shows that: (1) Spatially, there is obvious territoriality in the distribution of poverty-stricken villages, and the poverty-stricken villages are concentrated in contiguous poverty-stricken areas. The areas with the highest VPI, in a descending order, are Gansu, Yunnan, Guizhou, Guangxi, Hunan, Qinghai, Sichuan, and Xinjiang. (2) The main factors contributing to the poverty of poverty-stricken villages in rural China include road construction, terrain type, frequency of natural disasters, per capita net income, labor force ratio, and cultural quality of labor force. The main causes of poverty include underdeveloped road construction conditions, frequent natural disasters, low level of income, and labor conditions. (3) Chinese poverty-stricken villages include six main subtypes, and most poverty-stricken villages are affected by multiple poverty-forming factors, reflected by a relatively high proportion of the three-factor dominant type, four-factor coordinative type, and five-factor combinative type. (4) There exist significant poverty differences in terms of geographical location and policy support, and the governments still need to carry out targeted poverty alleviation measures according to local conditions. The research can not only draw a macro overall poverty-reduction outline of impoverished villages in China, but also depict the specific poverty characteristics of each village, helping the government departments of poverty alleviation at all levels to mobilize all kinds of anti-poverty resources.

Investigation of Optimal Segmentation Color Space of Bayer True Color Images with Multi-Objective Optimization Methods

For the selection of the optimal segmentation space of Bayer true color unmanned aerial vehicle image, this paper introduces multi-objectives constraints optimization to solve the inconsistency of multiple indicators. First, the Bayer color images were converted to YIQ(Luninance, Inphase, Quadrature), YCbCr(Luninance, Blue-difference, Red-difference), I1I2I3 (Three linear transformed color-opponent dimensions), HSI(Hue, Saturation, Intensity), Nrgb(Normalized Red, Green, Blue) and CIE(L*a*b*) (Comission Internationale de l′Eclairage, L*a*b* for Lightness and two color-opponent dimensions)color space, then the transformed images were segmented with multi-resolution segmentation method. By introducing the multi-objective constraint function, three parameters such as the topology index, geometric index and spectral area matching index were synthetically considered to determine the optimal segmentation scale. Based on that, the multi-objective constraint function was built to comprehensively analyze the result of segmentation, so as to find out the optimal color space for a certain type of building. And then the global optimum color space appropriate for all kinds of buildings can be gained through the comprehensive analysis of the F value of different types of buildings. Finally a series of images of different acquisition conditions and ground features were selected to conduct the test. The result shows that the optimal segmentation color spaces of different types of buildings vary a little. For cottage the I1I2I3 space can get the excellent object areas that reflect the real edge of the ground features, while the YCbCr space has some advantages on the segmentation of tile-building. Overall, only I1I2I3 color space has better integrated segmentation result for all buildings, and it is considered to be the best color space suitable for segmentation.

Graphics hardware-based algorithm for large scale terrain rendering

Used the main idea of Geometry Clipmap algorithm for reference, the emphasis focused on simplification of the algorithm and improvement of the details including LOD generation, data scheduling and data clipping etc. Designed a frustum shape mesh(View-Mesh), which was transformed into a terrain mesh by sampling the height map. Implemented the data clipping by using the current View-Mesh to clip a pile of DEM data arranged irregularly. And the terrain blocks selection and sampling was finished by GPU. Experiments show that this algorithm is a unified approach to process and render DEM data and texture data, it has a few advantages including simplicity of implementation, few data limitation and eliminating the cracks phenomenon effectively1.

Building extraction in towns and villages based on Digital Aerial Image by texture enhancing

With the continuous improvement of remote sensing image, especially the appearance of Digital Aerial Image, make the texture features of image more significantly and the deeply mining possibly and the object-oriented information extraction method has been applied on digital aerial image since the object-oriented information extraction method has been proposed, it has been applied in many fields and achieved better classification results than the traditional methods. In the process of object-oriented classification, the image quality of segmentation is the guarantee of the image classification accuracy. The current mainstream thinking of segmentation mainly considered about the four features of image, such as color, shape, smooth and compactness. The method based on texture enhancing of digital aerial image is brought on. In this article, firstly the original Digital Aerial Image is preprocessed by edge detection, principal component analysis and the texture filter of second-order probability statistics; Secondly the gray image of Contrast Texture was got through the sharpening window of 7×7; Then taking the gray image as independent band, false color composite were processed with the band combination(Contrast, the original Digital Aerial Image G, the original Digital Aerial Image B). The abundant of texture features in digital aerial image are translated into sensitive multi-scale segmentation spectral feature by using image enhancement technology, as it can promote segmentation effects. Building detail texture features have been involved in the process of multi-scale segmentation, and building segmentation will be more fully and sensitively. Finally, based on the false color image, multiple segmentations and building extraction in towns and villages were processed. Taking the Yanqing County Beijing Digital Aerial Image as an Example, building extraction was processed by method that mentioned above. Compared to the object-oriented classification method, it not only highlighted the edge of the buildings, but also reduced the redundant segmented objects. Besides it gets an effective solution to the shadow of the building and its confusing area, optimized the feature space, and improved the accuracy of classification.1

Three-dimensional display component for rural folk-custom tourism

China has a long history and rich culture landscape resources, but with not very well developed tourism. The importance of rural folk-custom tourism popularization and propaganda has brought attention and consensus to more and more people in modern society. Tourism publication is of great practical significance in tourism. With the problem that currently there were few systems about rural folk-custom tourism, we can only publish and query the character on tourism in the respective of directions of tourists and tourism resources, we rarely can intuitively publish, inquiry and analyze the interrelationship between the space characteristic information and the spatial information. So the consumers have no imagic and visual realization of the scenic spot, also lack the continuous display of the virtual reality space and interactive function. Considering this defect, we superpose the two-dimension electronic map to the digital elevation model to form the basic 3D simulation terrain of the demonstration zone for spatial analysis, integrated fast scheduling of the large-scale data, multi-scale level of detail of the landscape information, multimedia, WebGIS and virtual reality technology to develop the adaptive component, adaptive display the tourism landscape, for rural folk-custom tourism, for the interest of tourists, enterprise and the village administrative department. This paper indicates that the three-dimensional display component for rural folk-custom tourism will provide a new tool and verifying methods in tourism and it is extensible, portable and reusable, therefore, it simplifies the development and maintenance of applications.