Research On Buildings Extraction Of Remote Sensing Image Based On Morphology

With the development of society,remote sensing technology is gradually entering our field of view,in which the extraction of buildings in remote sensing images is regarded as an important subject of remote sensing,and with the maturity and progress of remote sensing technology,high-resolution remote sensing images provide a lot of help for our further research.Housing as the main body of the city,its extraction in urban planning,land detection and other fields have important applications,it is extremely influent and meaningful that how to efficiently and accurately extract the housing part of the image,for urban planning,precision agriculture and military detection.Based on the above research background and practical application,this paper puts forward a new method of extracting house objects through deeply research and experiments.The main work is as follows:(1)Preprocessing of remote sensing images.Aiming at the characteristics of remote sensing image,such as containing a huge quantity of data,radiation distortion,geometric distortion and multi-noise caused by various factors,the initial remote sensing image is preprocessed by image fusion,image denoising and image enhancement,obtaining a remote sensing image that is as realistic as possible in radiation and geometry.(2)Image segmentation based on edge detection.The traditional edge detection method is applied to our experiment,because of deficiency of its detection accuracy and anti-noise ability,it cannot meet our experimental needs.After several studies and experiments,choose the method of mathematical morphology to detect the edge of remote sensing image,and the traditional morphological edge detection method isimproved to construct a kind of edge detection Operator combining of multi-directional,multi-scale structural elements,which can preserve the original information of the image better while extracting the edge,after comparing with the experimental results of various edge detection methods,the experimental results are obviously better than those of other edge extraction methods.(3)Housing extraction and classification based on multi-feature.Firstly,the characteristics of the housing part of the remote sensing image are analyzed,including its geometric features,spectral features,texture features,etc.,for these features,we propose a method based on a combination of maximum correlation minimum redundancy(MRMR)feature selection algorithm and Fuzzy c-means(FCM).That is,according to these characteristics using MRMR feature selection method for the corresponding feature selection,and according to the preferred feature subset for FCM clustering,and finally we conduct further optimization process towards classification results map that we obtained.(4)Design comparative analysis experiment.Selecting sample data,adopting different feature selection and classification methods to process extraction experiment to selected remote sensing housing image,and the classification results of different methods were analyzed in precision evaluation by using the confusion matrix.The results show that the method adopted in this paper can be used to extract the housing information quickly and accurately,which can be used as an effective method for extracting houses from remote sensing images.