Research On Remote Sensing Image Segmentation Of Building Based On Morphology

The traditional remote sensing image building measurement method is mainly based on manual visual interpretation.This manual measurement method is time-consuming and labor-intensive,and the cost is high,the measurement result is not ideal.With the implementation of the major special projects of the high-resolution Earth observation system,the GF-1 and GF-2 satellites have been successfully launched successively,and the image analysis method can be used to identify the building targets in the remote sensing image.On the one hand,it is convenient for urban planning management,geographic database updating and decisions;On the other hand,it provides data support for relevant departments.Because the remote sensing image is relatively complex,the phenomenon of mis-segmentation may occur when extracting the building area.The morphological method can be used to extract the target to keep the target contour position unchanged.Therefore,based on the morphological theory method,this paper proposes a corresponding segmentation method for the remote sensing image building area.The main work of the thesis is as follows:(1)Firstly the current situation of building extraction research at home and abroad is analyzed for remote sensing image building segmentation,Then the basic theory and method of mathematical morphology for remote sensing image processing is analyzed,and relevant simulation experiments and comparisons are made.(2)There are problems such as incomplete edges and poor noise suppression when a single fixed morphological structuring element is used to detect the edges in remote sensing image building.For this reason,a morphological edge detection method for remote sensing image building based on variable structuring element is proposed.Firstly structural elements with different scales and multiple orientations are constructed to make corresponding morphological operations.In order to suppress the noise in the target background,highlight the edge of the image of the building,the constructed morphological operation is used to preprocess the remote sensing image;Then adaptive morphological edge detection is used to obtain multiple images with different scales and directional edge features;Finally the building image edges are obtained by weighted summation of each direction edge,and then the least square is used to fit the edges for accurate location of the edge contour of the building target.The comparison of experimental results show that the proposed method can detect complete building edge information,with high edge detection accuracy and superior anti-noise performance.(3)Aiming at the low contrast,and the indistinct distinction between target and background in some remote sensing images,a morphological remote sensing image building segmentation method based on GAN adaptive structural elements is proposed.Firstly the pixel of the building area to be processed is selected as the seed point pixel,and the adaptive neighborhood is defined according to the brightness feature of the image pixel and the neighborhood constraint relationship,Thus the constructed the adaptive neighborhood as the adaptive structural element;Then the morphological basic operations of the corresponding adaptive variable structure elements are constructed to enhance the remote sensing image;Finally the enhanced remote sensing image is segmented by the grayscale slice method to segment the building area.The experimental results show that the proposed method can accurately locate the target edge contour while segmenting the interior of the building image target,and has superior anti-noise performance.Compared with the classical threshold segmentation and region segmentation algorithm,the segmentation accuracy of this paper can reach 98%,and the over-segmentation rate can be controlled within 3%.