Super-Resolution Reconstruction Method for Single Space Object Image based on Optimized Convolution Neural Network

To date, a large number of platforms have been launched into outer space. More and more attention has been paid to space object images. However, it is difficult to obtain high-quality spatial images. One of the most important reasons is that the sensor technology is not good enough. Inspired by the typical super-resolution reconstruction method based on deep CNN, a super-resolution reconstruction method of single space target image based on deep CNN is proposed. This method uses Nesterov acceleration gradient method to train the network. The experimental results show that the method is feasible in the reconstruction of high-resolution spatial target image.