Design And Implementation Of Remote Desktop Based On Neural Network Codec

Remote desktop is a software that provides remote control functions.Through remote desktop,you can connect to a remote computer,access all its applications,files,and network resources,and realize real-time operations.With the rapid development of Internet technology,the continuous advancement of 5G technology and the cloud With the rapid development of computing technology,remote desktop technology will be applied in more fields.Thanks to the rapid development of 5G networks,the network transmission speed has been qualitatively improved.Therefore,remote desktop technology can be applied to real-time remote services,such as remote maintenance,remote office,remote monitoring and management,and remote assistance.The rapid development of technology,cloud computing technology,artificial intelligence and other technologies.Remote desktop will not only make rapid progress in traditional application fields,but also combine it with many advanced technologies now to bring more convenient and efficient services,such as remote desktop technology and cloud The combination of service technologies can provide virtual computer services and remote maintenance services.Although the development of Internet technology and network communication technology has provided good external conditions for remote desktop technology,the optimization of internal components and operating mechanisms of remote desktop technology will also promote the progress of remote desktop technology.The implementation mechanism of the remote desktop software is to implement the remote desktop control of the computer based on the Internet and computer technologies such as computer network communication protocols,picture encoding and decoding technology,and computer frame buffer technology.Among them,the SPICE protocol,VNC protocol and RDP protocol are currently the most mature.This paper uses the VNC protocol implementation software tigerVNC as the optimization software to achieve the remote desktop software with higher resolution and faster response speed by implementing the optimized VNC software picture encoding and decoding technology.At present,with the rapid development of big data technology and neural network technology,machine learning technology is rapidly maturing,and the combination of machine learning technology with traditional computer and Internet technology has made traditional technology significantly improved and improved.Deep learning is used to complete the optimization of traditional picture coding and decoding,and once again based on multi-layer convolutional neural network based picture coding and decoding model to optimize tigerVNC software.The RFB protocol of the remote desktop provides a variety of lossless and lossy encoding and decoding schemes,but currently the learning-based image compression technology has developed rapidly,and the encoding and decoding effects have been higher than traditional image compression technologies such as jpeg and web p.Therefore,the remote desktop technology has passed A picture codec based on a multi-layer convolutional neural network can get better results.This project uses the open source project tigerVNCServer as the remote desktop VNCServer,uses tigerVNCViewer as the VNCViewer,uses tensorflow to construct a multi-layer convolutional neural network plus quantization and entropy coding to form an encoder model,and constructs a multi-layer convolutional neural network plus entropy decoding.Make up the decoder model.The codec based on neural network is embedded in tigerVNC to obtain the optimized remote desktop model.