An Environmental Characterization Method for Monocular Vision Obstacle of Rotor Aircraft

Abstract

Abstract Combining the structure of coding and anti-coding in the field of deep learning, we explore a method to represent the complex environment of four rotor aircraft. First, we use this structure to explore the process of dimensionality reduction and reconstruction of high dimensional image data. Second, combining this structure, we use the concept of transposition convolution to propose a network structure for depth distance estimation using monocular vision images. Finally, we get the direct physical feature information used to describe complex environment.