Space-based Atmospheric Background Image Data Classification Method And Its Characteristic Analysis

The radiation characteristics measurement of the atmospheric background provides a great deal of data support for the parameters of the space-based early warning system.It is a basic work for the construction of the space-based early warning system.In recent years,the spatial image data grows in the number of terabytes with the development of new detectors.As a result,It becomes prerequisites and basis for the statistic of atmospheric background radiation characteristic to store and manage spatial data effectively and dig information from it.This article is based on the measurement data from a atmospheric background measurement satellite of China.The purpose of the measured radiation characteristics is to provide infrared radiation based on the background simulation data and it carries out the research mainly on storage and management of atmospheric background image data,classification of atmospheric background image and statistical and analysis of radiation data characteristic of atmospheric background image.In terms of storage and management of atmospheric background image data,it elaborates the data model and organization method of the spatial data.Based on that and the comparison of storage methods,the Geo Raster component of the Oracle 11 g is used as the database for the storage of atmospheric background image data.Meanwhile,it designs the storage structure of the infrared radiation database of the atmospheric background and imports the pixel and attributes information of the image to the image library based on the actual image information.Besides,based on the analysis of the location index and spatial relationship of the image,the images in the library are geographically registered and allocated with specific spatial indexes.In terms of classification of atmospheric background image,based on the classification data set,it elaborates the principle of the Convolutional Neural Network and establishes two CNN model to divide the images into two categories: Earth's atmospheric background images and subterranean atmospheric background images.Besides,it also trains the classification model and tests the accuracy based on the classification training and test samples.In terms of analysis of radiation data characteristic of atmospheric background image,according to the classified two image categories,it sets up the statistic tables of the Earth's background infrared radiation and subterranean background infrared radiation,respectively and simulates the diagrams of atmospheric background infrared radiation and the variation of the radiance value with the height of the tangent on the basis of the statistical results.Finally,it carries out some simple analysis based on the results above.