Research And Application Of Deep Learning In Video Interframe Compensation

In the transmission of video data in the network,due to the delay of the network line,the congestion may lead to the loss of data packets,which leads to the poor quality of the video picture decoded by the receiver.In this paper,the problem is deeply studied,and an algorithm of video inter-frame compensation based on depth learning is designed: an unsupervised neural network model for video image reconstruction.Based on the understanding of the details of the whole video content structure,the missing part of the video frame is reconstructed.Experimental results show that the proposed neural network model is more effective than the algorithm based on sample interpolation in video inter-frame compensation.The main contents of this article are as follows:1.This paper studies the development,prospect and various application scenarios of fusion communication,studies the strong demand of communication system for network video transmission,and studies and analyzes the advantages of the combination of network video super-resolution reconstruction and machine learning.2.The related principles of deep learning,common models,convolution neural networks and so on are studied.The basic principles of TensorFlow,a popular open source framework for deep learning,are studied in combination with the theory of depth learning.The image recognition algorithm based on depth learning is improved and compared with other image recognition algorithms.The results show that the improved algorithm improves the accuracy of image recognition.3.This paper studies video coding and decoding,and gives two important reasons for video coding at present: first,the video files are generally large and need a lot of storage space to save all kinds of video files in life.Secondly,because the video file is relatively large,the video data takes up most of the bandwidth in the network transmission,which leads to the network congestion.Therefore,the transmission of video data will bring great pressure to the network.Occupying large bandwidth eventually leads to network delay and serious data frame loss.The result is that the video picture of the receiving end is not clear or the video file is destroyed.Many scholars have made great efforts to solve these two problems.They have made remarkable achievements in the field of video coding and video compression.Now H.264 is one of the mainstream coding standards.This paper makes a deep research on H.264 video coding and decoding framework.4.Convolutional neural network has inherent advantages in image processing.In this paper,a convolution neural network model for video frame compensation is designed by using deep learning to deal with video image problems.The neural network model is compared with other algorithms,and the results show that the proposed algorithm is effective.