| The rapid development of remote sensing technology in China in recent years,optical remote sensing satellite image resolution continues to improve,such as ZY series,Jilin series,Super View series,GF series satellite launch followed by a dense,remote sensing data showing explosive growth.At present,most remote sensing images processing and analysis are implemented by professional personnel on local machine.The processing efficiency is low and can not meet the needs of mass remote sensing images processing.Moreover,from remote sensing data to thematic data products,a series of processing steps are needed,and different software tools may be applied to different requirements according to different requirements.A great decrease in timeliness.With the rapid development of network technology and the wide application of cloud computing,artificial intelligence,deep learning,big data and so on,how to quickly provide users with remote sensing data,computing resources,algorithm model resources and their special products by online processing has become a difficult and hot issue.This paper is a high resolution.The online remote sensing images processing platform is researched and implemented,and integrating remote sensing data processing algorithms and processing tools resources.The main research work of this paper is as follows:(1)The technical basis and service development mode of the existing remote sensing images online service are introduced in detail.The custom professional remote sensing images processing model is deployed on the cloud,and the client is called through the RESTful API interface.Based on the MEAN development framework,a high resolution remote sensing image online processing platform is designed and implemented.A kind of online,real time,convenient and professional service resources,it realizes the one-stop operation of the shopping cart,such as image query,remote sensing images coverage statistics,image orthophoto correction fusion,information extraction,result browsing,downloading and so on.(2).The process of making and service publishing based on domestic multi source satellite image orthophoto correction and fusion is studied,and the one key orthophoto correction and fusion operation is implemented and it is published as a service.At the same time,in order to distinguish the existing mature commercial application framework,such as ENVI Service Engine,Digital Globe GBDX remote sensing big data platform,this paper proposes docker container technology to encapsulate third party remote sensing images processing algorithm model and realize the RESTful architecture style Web Service solution.(3)Based on the research of remote sensing image processing algorithms for remote sensing image online processing platform,a remote sensing image classification algorithm based on full convolution neural network is proposed.After the convolution and pool layer,the fusion layer and the deconvolution layer are added to reduce the loss of image feature.And the Mean-shift algorithm is used to get the spatial relationship between pixels to optimize the classify result in the edge region. |
PDF Full Text Request