Target And Environment Characteristics Analysis Of Remote Sensing Image

Remote sensing images could be seen as the perceptional results of surface featuresand scenes from a great distance, and the targets and environment on the image contain awealth of information. As the interested object to detect and identify, analyzing targetcharacteristics is a main content of the remote sensing image processing. However, theeffect of complex environment on target characteristics appears increasingly prominent inthe continuous development of quantitative remote sensing. Consequently, environmentalcharacteristics also need to be considered in the analysis of target characteristics.Introducing environment information, focusing on target from its environment, usingenvironmental characteristics to serve as a foil to target characteristics, which is calledenvironmental perception technology, has considerable theoretical significance andpractical value in remote sensing image analysis for military applications.In study of the target characteristics analysis of remote sensing image, based onhyperspectral image atmospheric radiation correction, this paper discusses theatmospheric environment effect on the spectral characteristics, grayscale characteristicsof the target in remote sensing image and common parameters sensitive to targetcharacteristics, while also makes a general introduction to the atmospheric environmentcharacteristics. Combining of actual project applications, we should consider the impactof complex environment on robustness in the scheme demonstration and algorithmoptimization.In study of the environment characteristics analysis of remote sensing image, thispaper uses cloud as a special case of complex environment whose spectral and texturefeatures are analyzed in detail. On this basis, we put forward a tree-like cloud detectionalgorithm based on multi-cloud characteristics taking the spectral and texture informationinto account at the same time, which test results are broken down into four types, thickcloud, thin cloud, probable cloud and clear area. Its greatly improved accuracy has beenverified by comparing with basic algorithms through simulated image. As for the imagewith thin cloud, this paper proposes an improved displacement algorithm based on cloudfeatures to remove the thin cloud, and the algorithm achieves satisfactory results in thevisual evaluation and quantitative analysis by experiments.Aiming at specific applications-oriented practical requirements, while consideringassociation of target and environment characteristics, this paper innovatively introducesthe environmental perception conception into the field of remote sensing imageprocessing, and puts forward a target characteristics analysis and recognition technology based on environmental perception, with building a typical scene environmentknowledge base. The experimental results show the environmental perception technologycan effectively overcome the drawbacks of traditional target recognition which onlyconsidering the target characteristics can not distinguish false target from the true ones inparticular cases, and solve the low rate of target recognition problem when under therestrictions of lacking a priori knowledge.