A Deep Image Denoising Method at Transmit Electricity Surveillance Environment.

Deep neural network has extraordinary advantages in the recognition and restoration of background pictures of transmit electricity surveillance environment. However, the existing network framework for identifying pictures and removing noise in the background pictures of transmit electricity surveillance environment still has very big problems, so our current challenge is to design a network framework suitable for identifying transmit electricity surveillance environment scenes and removing picture noise in the t transmit electricity surveillance environment. At present, most methods based on depth neural network are directly based on making poor quality images into good quality images to solve the problem of background pictures of transmit electricity surveillance environment, but this process will lead to the observation model that ignores the image degradation process, resulting in the difference between the final recognized image and our expected effect. In this paper, we make a deep image denoising algorithm about background pictures of transmit electricity surveillance environment. We can effectively calculate the iterative steps through the algorithm. So we expand the iteration into a DNN(deep neural network) on identifying the transmit electricity surveillance environment. The denoiser can make the most of the multi-scale redundancy of transmission line pictures. Therefore, we proposed network uses the excellent ability of deep neural network to remove image noise, but also uses the prior knowledge of image model under the background of power line. Through the node-to-node optimization training iteration, the network of different modules can be combined for optimization. The experimental results of several image recognition and optimization tasks in background pictures of transmit electricity surveillance environment show that our method can achieve very excellent experimental results.