水下
计算机科学
人工智能
计算机视觉
遥感
噪音(视频)
地质学
声学
跟踪(教育)
特征(语言学)
出处
期刊:
[SPIE]
日期:2026-05-14
卷期号:: 3-3
摘要
Shallow-water environments are critical areas for underwater security, where small underwater vehicles play a vital role in maritime operations. One of their most significant characteristics is their stealth capability. However, due to the complexity of ambient noise and irrelevant signals in underwater environments, accurate target recognition remains challenging, and the recognition rate is often affected. In this study, we propose a method for identifying underwater vehicles of different materials and shapes using linear frequency modulation (LFM) signals transmitted by active sonar. First, the received signals are preprocessed using the Fast Independent Component Analysis (Fast-ICA) algorithm. Then, wavelet transform is applied to further denoise the echo signals, enhancing the signal-to-noise ratio by separating the signal from noise. Next, a combination of Short-Time Fourier Transform (STFT) and Hough Transform is employed for feature extraction from the sonar echoes. Finally, a Support Vector Machine (SVM) classifier is used to recognize and classify the underwater vehicles.
科研通智能强力驱动
Strongly Powered by AbleSci AI