Paper
27 March 2022 Distance estimation of obstacles in monocular vision based on deep learning
Author Affiliations +
Proceedings Volume 12169, Eighth Symposium on Novel Photoelectronic Detection Technology and Applications; 12169B3 (2022) https://doi.org/10.1117/12.2626519
Event: Eighth Symposium on Novel Photoelectronic Detection Technology and Applications, 2021, Kunming, China
Abstract
Aiming at the forward-looking distance perception requirements of obstacles such as mountains, forests and trees during the ultra-low-altitude flight of subsonic aircraft, an obstacle distance estimation method based on deep learning is proposed. Firstly, the relationship between adjacent hyper-pixels is modeled by the conditional random fields, the neural network model is trained using the data set, and then the loss function of the pose parameter regression is designed, and the representation of the neural network model is input to the constructed loss function to optimize the weight of the regression model. Simulation experiments show that the proposed method can effectively perceive obstacles within 3km ahead, and the distance estimation error is not more than 20% of the relative distance, which can meet the obstacle avoidance requirements of most sudden obstacles during ultra-low-altitude flight.
© (2022) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Shan Liu, Long Ming Li, and Dong Hui Wei "Distance estimation of obstacles in monocular vision based on deep learning", Proc. SPIE 12169, Eighth Symposium on Novel Photoelectronic Detection Technology and Applications, 12169B3 (27 March 2022); https://doi.org/10.1117/12.2626519
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Data modeling

Visual process modeling

Visualization

Computer simulations

Image processing

Forward looking infrared

Image analysis

Back to Top