딥러닝을 이용한 DEMON 그램 주파수선 추출 기법 연구A study on DEMONgram frequency line extraction method using deep learning
- Other Titles
- A study on DEMONgram frequency line extraction method using deep learning
- Authors
- 신원식; 권혁종; 설호석; 신원; 고현석; 송택렬; 김다솔; 최강훈; 최지웅
- Issue Date
- Jan-2024
- Publisher
- 한국음향학회
- Keywords
- Deep learning; Semantic segmentation; Detection of Envelope Modulation on Noise (DEMON); DEMONgram; Frequency line extraction; 딥러닝; 의미론적 분할; Detection of Envelope Modulation on Noise (DEMON); DEMON 그램; 주파수선 추출
- Citation
- the Journal of the Acoustical Society of Korea, v.43, no.1, pp 78 - 88
- Pages
- 11
- Indexed
- SCOPUS
ESCI
KCI
- Journal Title
- the Journal of the Acoustical Society of Korea
- Volume
- 43
- Number
- 1
- Start Page
- 78
- End Page
- 88
- URI
- https://scholarworks.bwise.kr/erica/handle/2021.sw.erica/118463
- DOI
- 10.7776/ASK.2024.43.1.078
- ISSN
- 1225-4428
2287-3775
- Abstract
- 수중 소음 측정이 가능한 수동 소나에 수신된 선박 방사소음은 Detection of Envelope Modulation on Noise(DEMON) 분석으로 얻은 선박 정보를 사용하여 선박 식별과 분류가 가능하다. 하지만 낮은 신호대잡음비(Signal-to-Noise Ratio, SNR) 환경에서는 DEMON 그램 내 선박 정보가 담겨있는 표적 주파수선을 분석 및 파악하는데 어려움이 발생한다. 본 논문에서는 낮은 SNR 환경에서 보다 정확한 표적 식별을 위해 딥러닝 기법 중 의미론적 분할을 사용하여 표적 주파수선들을 추출하는 연구를 수행하였다. SNR과 기본 주파수를 변경시키며 생성한 모의 DEMON 그램 데이터를 사용하여 의미론적 분할 모델인 U-Net, UNet++, DeepLabv3+를 학습 후 평가하였고, 학습된 모델들을 이용하여 캐나다 조지아 해협에서 측정한 선박 방사소음 데이터셋인 DeepShip으로 제작한 DEMON 그램 예측 성능을 비교하였다. 모의 DEMON 그램으로 학습된 모델을 평가한 결과 U-Net이 성능이 가장 높았으며, DeepShip으로 만든 DEMON 그램의 표적 주파수선을 어느 정도 추출할 수 있는 것을 확인하였다.
Ship-radiated noise received by passive sonar that can measure underwater noise can be identified and classified ship using Detection of Envelope Modulation on Noise (DEMON) analysis. However, in a low Signal-to-Noise Ratio (SNR) environment, it is difficult to analyze and identify the target frequency line containing ship information in the DEMONgram. In this paper, we conducted a study to extract target frequency lines using semantic segmentation among deep learning techniques for more accurate target identification in a low SNR environment. The semantic segmentation models U-Net, UNet++, and DeepLabv3+ were trained and evaluated using simulated DEMONgram data generated by changing SNR and fundamental frequency, and the DEMONgram prediction performance of DeepShip, a dataset of ship-radiated noise recordings on the strait of Georgia in Canada, was compared using the trained models. As a result of evaluating the trained model with the simulated DEMONgram, it was confirmed that U-Net had the highest performance and that it was possible to extract the target frequency line of the DEMONgram made by DeepShip to some extent.
- Files in This Item
-
Go to Link
- Appears in
Collections - COLLEGE OF SCIENCE AND CONVERGENCE TECHNOLOGY > DEPARTMENT OF MARINE SCIENCE AND CONVERGENCE ENGINEERING > 1. Journal Articles
- COLLEGE OF ENGINEERING SCIENCES > SCHOOL OF ELECTRICAL ENGINEERING > 1. Journal Articles

Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.