A Joint Underwater Target Localization and Recognition System Based on Bionic Active electrosensing and Deep Learning
Meijiang Hou, Jiegang Peng
University of Electronic Science and Technology of China
阅读操作
确认中在文库中上传 PDF 后可生成中文音频讲解。
摘要与影响
Inspired by the active electroreception mechanism of weakly electric fish, this paper proposes a joint underwater target localization and recognition system that integrates bio-inspired electric sensing with deep learning. The system employs a dipole-like electrode array to establish a quasi-static electric field and uses a 2D receiving array to acquire voltage signals induced by the target. The voltage amplitude curves are treated as image-like signals, and a large dataset is collected from 24 types of targets in a controlled 3D space. A 2D convolutional neural network is used for end-to-end feature learning, simulating the biological adaptive learning process to simultaneously train a position/size regression model and a material/shape classification model. Real-time field test results show that the system performs robustly in multi-parameter analysis: the mean absolute errors (MAEs) for x, y, and z localization are 3.823 cm, 2.941 cm, and 2.876 cm, respectively; the length prediction MAE is 0.321 cm; and the classification accuracies for material and shape reach 83.8% and 85.6%, respectively. The system demonstrates high accuracy and strong generalization ability under controlled environments. This study reveals the innovative potential of combining bio-inspired perception with deep learning, providing a new paradigm for anti-interference, multi-parameter underwater detection.
逐年被引趋势
暂无年度引用数据
关键指标
同类平均 = 1
同领域 · 同年份 · 同类型
Google Scholar 与 OpenAlex 的被引统计范围不同,数值存在差异属正常。
AI 辅助阅读
依据:摘要
可就本文提问;依据不足时会说明。
学术脉络
学科主题
物理Underwater Acoustics Research
Water Quality Monitoring Technologies · Underwater Vehicles and Communication Systems