Forward denoising neural network for underground mine images

Delin An; Huping An; Shaozheng An; Zhimei Zhang

doi:10.1117/12.2628415

21 April 2022 Forward denoising neural network for underground mine images

Delin An, Huping An, Shaozheng An, Zhimei Zhang

Proceedings Volume 12175, International Conference on Network Communication and Information Security (ICNCIS 2021); 1217509 (2022) https://doi.org/10.1117/12.2628415
Event: International Conference on Network Communication and Information Security (ICNCIS 2021), 2021, Beijing, China

Abstract

Coal resources are one of the vital energy sources in China. To ensure the safety of mine tunnels, drilling anchor robots are widely used in the task of anchor net support of mine walls. Since the anchor drilling robot needs to accurately identify and locate anchor holes and then install support anchors under harsh conditions such as dust and water mist, it needs to obtain clear images of the underground. This paper designed a forward-perceptive neural network model based on a convolutional neural network and attention-based layer structure to solve this problem. The model enhances the noise reduction capability of classical neural networks by forward perceptive design and uses pixel-level perceptual neurons to ensure that the distribution of effective pixel points is not affected during the denoising process. Finally, the model is tested to process the noisy underground mine images collected in the field. The signal-to-noise ratio of the processed images reaches more than 30, which exceeds the traditional noise reduction method by 1.3 times. Meanwhile, the pixel distribution of the processed images is not distorted.

Citation Download Citation

Delin An, Huping An, Shaozheng An, and Zhimei Zhang "Forward denoising neural network for underground mine images", Proc. SPIE 12175, International Conference on Network Communication and Information Security (ICNCIS 2021), 1217509 (21 April 2022); https://doi.org/10.1117/12.2628415

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available