| Satellite remote sensing observation is of great significance to the management and planning of the environment.Compared with the ground observation,satellite remote sensing observation has the advantages of wide coverage area,continuous observation time and sustainable dynamic observation.Satellite remote sensing can be divided into polar satellite remote sensing and still satellite remote sensing,polar satellites with day-to-day cycles to limit the time resolution,can not be real-time observation,while the static satellite has a higher time resolution,The new generation of stationary satellites lifted the aerial satellite to improve the observing ability,compared with the traditional static satellites,the new generation of stationary satellites have more spectral channels and higher temporal and spatial resolution.The new generation of still satellites Himawari8 used in this paper has more spectral channels,higher temporal and spatial resolution than conventional geostationary satellites,according to the characteristics of Himawari8 cloud-snow detection and surface reflectivity library to build the effect and the use of traditional satellite for the effect of such algorithms is also better(1)Research on algorithm of cloud-snow detectionCloud cover is the most common obstacle of remote sensing image processing,and snow has the characteristics of high reflectivity,which directly affects the accuracy of aerosol inversion.In order to improve the accuracy of inversion products,cloud snow detection is very important in remote sensing image preprocessing.In this paper,we discuss the algorithm of snow detection using the new generation of stationary satellites Himawari8.The algorithm makes full use of Himawari8’s characteristics of 16 spectral channels,synthesize the complementary characteristics of visible and thermal infrared channels,and perform multi-spectral detection method.Cloud and thick cloud recognition,at the same time using the exponential method to achieve the identification of snow and ice.Finally,the results of the algorithm are compared with the original data.The results show that the cloud snow detection of this algorithm has a good effect.(2)Research on algorithm of surface reflectance library constructionBased on the advantages of high time resolution and high spatial resolution of the still satellite Himawari8,the surface reflectivity library of the satellite is constructed.The surface reflectivity library is a time series based image set with different wavelengths.The surface reflectance is Many of the important variables of the surface parameters.In this paper,the Himawari8 data has radiometric calibrated at first,the solar zenith angle is extracted from the remote sensing image,the auxiliary information such as the zenith angle is observed,and the surface reflectance of each pixel is calculated by dark pixel method.After the construction of the surface reflectivity library.Finally,the experimental results show that the NDVI and NDVI of the vegetation are increased,and the validity of the surface reflectance is proved by calculating the NDVI.(3)Realization of parallel acceleration of algorithm of cloud-snow detection and algorithm of surface reflectance library construction by GPUThis algorithm is based on a new generation of stationary satellites,which have multiple spectral channels,high temporal and high spatial resolution characteristics,which result in massive remote sensing data.In order to meet the requirement of remote sensing image in the practical application,it is very important to improve the processing speed of remote sensing image.In this paper,the CPU+GPU heterogeneous system,combined with the CPU and GPU different computing power,a reasonable division of computing tasks,and the realization of the host-device side,arithmetic instruction flow,the use of texture memory and constant memory and other aspects of optimization,And the acceleration of the surface reflectivity library construction algorithm.Finally,through quantitative experiments,the experimental results show that the effectiveness of GPU acceleration. |
PDF Full Text Request