Study Research On Extraction Of Spatial Information Of The Edge Of The Building Based On High Spatial Resolution Images

With the continuous development of urbanization process, artificial feature and urban population, economy and other factors are closely related. The artificial features obtained quickly and accurately from the remote sensing images are not only useful to geographical spatial data updating, but also important to the effective monitoring of new building information. High resolution remote sensing data can provide richer detail information, which is helpful to recognize and classify artificial features. Therefore, this article explores building extraction based on high resolution remote sensing images from the following aspects:Firstly, the paper introduces the background of high resolution remote sensing image’s development, and then summarizes the domestic and foreign research status in related fields. From the aspects of remote sensing development, high resolution remote sensing data development and remote sensing technology development, it introduces the general trends of remote sensing technology, then elaborates the main application of high resolution remote sensing in city level, which includes extraction and investigating detection of artificial features. Besides, it introduces the main content of this article and technical route.Secondly, the paper introduces the characteristics of the study area and WorldView-2data, analyzes the characteristics of several typical object and spectral characteristics in original data with statistical methods, and evaluates the data quality after data fusion. We also introduct the characteristics of the building.Thirdly, the paper explores the method of building edge information extraction from high resolution remote sensing images. Using classical operators such as Roberts Prewitt, Sobel, LapLacian, Kirsch operators, and second operator such as the Laplasse Gauss operator, Canny operator, and the wavelet transform, Gauss Laplasse pyramid method in the C#.net2010platform, we experiment the building edges extraction on the worldview remote sensing image data. We put forward a edge detection method with adaptive canny wavelet and dual threshold selection. This method can maintain the weak edges. It has a higher recognition rate and accuracy. The detected edge is smooth and delicate, with less noise point.It has great significance in image processing.Finally, using supervised classification methods like neural network classification and support vector machine classification, we extract building information from the image, and vectorize the building classification image. With post-processing in ARCGIS, we get the shape building vector data, so as to get the building edge information, and compare with edge detection method which combined to Canny and wavelet adaptive double threshold. We found that the combination of spatial information building edge detection method and wavelet adaptive canny choice of dual-threshold obtained is linear edge building spatial information, can reflect more intuitive edge features of buildings, with high accuracy. The method of classification of images obtained from high spatial resolution remote sensing is a planar edge of building, with more localized error。...