Design And Implementation Of Remote Sensing Image Processing System Based On Deep Learning

Remote sensing image target recognition technology is widely used in the fields of earth observation,mineral resource management,surveying and mapping,and military target detection.At present,the proposed remote sensing image target recognition algorithm is too targeted and the accuracy of the algorithm needs to be improved.Therefore,the remote sensing image target recognition algorithm needs further research.Based on the high-resolution remote sensing image data,this paper uses the deep learning image processing technology to introduce the instance segmentation technology into the remote sensing image target recognition,and constructs a remote sensing image target segmentation recognition network based on convolutional neural network.At the same time,a remote sensing image processing system based on deep learning is realized.The specific work of this paper includes the following three points:(1)A deep learning network based on Mask R-CNN is constructed for the recognition and segmentation of remote sensing image targets.The method uses the dense convolution network and the feature pyramid network to extract the features of the image,and the obtained feature maps of different scales are input into th e region generation network to obtain candidate frames containing the targets,and the candidate frames are scaled by RoIAlign,and then in each Class determination,rectangular frame correction,and instance segmentation are performed on the candidate regions.For the segmentation task in the network,an 18-layer network' structure based on' the SegNet model is constructed,and the segmentation effect is improved.Finally,the constructed network model is trained and tested by using the publicly available remote sensing image house dataset and manually annotated aircraft dataset.The test results show that the method used in this paper can effectively improve the accuracy of remote sensing image target recognition and can be used.Remote sensing image target recognition and segmentation field.(2)A remote sensing image processing system based on deep learning was designed and implemented.The built-in deep learning model,combined with Django,RabbitMQ and Celery technologies,includes four modules:dataset management,model management,task management and user management,which respectively implement dataset upload,dataset display,model upload,Model display,model deletion,model download,model training,identification of uploaded remote sensing images and user management.(3)The remote sensing image processing system was tested.Functional testing of the system is performed by using the black box test method to simulate the user's operational behavior.Performance analysis of the system based on the test results.The test results show that the system achieves the expected effect and can be used to identify and segment remote sensing image targets.