Drone Image Reconstruction Using Autoencoder Architecture

The work is devoted to the process of investigation and development of the approach to the reconstruction of images taken from unmanned aerial vehicles (drones) for further work with them. Proposal of the approach is based on using neural networks with an auto encoder architecture, which allows us to reconstruct images with high accuracy. Experiments have shown that the mapped architecture demonstrates a high level of accuracy and brightness of the updated image. The results of the work can be used for developing processing systems for drones or other applications.