Research On Information Extraction From High Resolution Remotely Sensed Imagery

High resolution remote sensing image data has become the basic resource of national spatial data infrastructure and strategic layout of various countries as a spatial data carrier that has large amounts of information,extensive coverage,and is economical and convenient,which has the feature of obtaining surface information without touching with ground object target.Although spatial imaging resolution is improved gradually,it is still an important scientific problem to be resolved at present about how to efficiently and automatically process and extract information of image data towards the remote sensing image that has big data and complex content.Therefore,this paper aims to study the interpretation extraction method of the ground object information of water and residents that is applied to high-resolution remote-sensing image,so as to resolve the insufficient processing capacity of remote sensing image data and the extraction precision difference,low efficiency and other problems caused by the application of traditional moderate and low resolution remote sensing image extraction method during interpreting high-resolution image.The main research achievements of this paper are as follows:(1)Introduces the development process and significance of high-resolution satellite remote sensing technology;summarizes the current main analysis methods of remote image;and analyzes the research status of typical ground object information extraction of water and residents against the research content of this paper,and makes the research scope and basic idea clear.(2)Improves the fuzzy connectedness algorithm,and makes it apply to water information extraction of high resolution remote image,and designs the local fuzzy connectedness algorithm through defining and proving the significance of valid path,invalid path and fuzzy neighborhood connectedness under the theoretical system of fuzzy connectedness,and can completely extract the water information of full image.(3)The filter under the homomorphic system was designed to finish the extraction of river information.The proposed algorithm used spectral analysis of remote sensing image to define the feature identification of river information in the frequency domain.The one-dimensional profile line adding-windows analysis river location technology was adopted to realize river location and width estimation of the remote sensing image automatically and quickly.Finally,the low-pass filter under the homomorphic system was designed to extract low frequency river information.(4)Combination of directional features of texture information in residents and Harris corner detection,proposes a information extraction method in settlement place of high resolution remote sensing image;first,conduct filtering transformation at specific direction and angle to experimental image through Gabor filter;next,conduct Harris corner detection calculation likelihood function to filtering response figure;finally,extract the information of residents on the image according to the method of threshold segmentation.