基于自适应空间特征增强的轻量级声呐图像目标检测
陈启北,史永桥,韩路军,陈慧
摘要(Abstract)：
声呐图像处理和分析在水下目标自动检测中起着至关重要的作用。然而，声呐图像低清晰度、大噪声的物理属性严重影响了目标检测模型的识别精度。此外声呐图像的采集困难、样本量较少等因素也给声呐目标的探测精度带来了巨大的挑战。针对这些问题，提出了一种基于自适应空间特征增强的轻量级声呐图像目标检测模型，发明了一种轻量化CSPDarkent；提出一种多尺度残差块，提升特征的多梯度流信息和模型的回归检测精度；最后，使用改进空间坐标卷积替换原始2D-Conv，增加特征的空间维度信息，提升模型对目标的关注度。实验的结果表明，该模型算法在检测精度和检测速度上均超过现有方法，实现了两者的有效平衡，为后续的声呐图像目标检测模型的设计提供新的解决思路。
关键词(KeyWords)： 空间特征融合;声呐图像目标检测;YOLOv5;特征增强
基金项目(Foundation): 广西研究生教育创新计划项目——“基于人工智能和协同蜜罐集群的海洋信息系统智能诱导安全防御体系的研究”,项目编号：YCSW2022289
作者(Author): 陈启北,史永桥,韩路军,陈慧
参考文献(References)：
- [1]Villar S A, Acosta G G, Solari F J. OS-CFAR process in 2-D for object segmentation from Sidescan Sonar data[C]//2015 XVI Workshop on Information Processing and Control(RPIC). IEEE, 2015:1-6.
- [2]YAN X, LI J, HE Z. Measurement of the echo reduction for underwater acoustic passive materials by using the time reversal technique[J]. Chinese Journal of Acoustics, 2016, 35(3):309-320.
- [3]Raghuvanshi D S, Dutta I, Vaidya R J. Design and analysis of a novel sonar-based obstacle-avoidance system for the visually impaired and unmanned systems[C]//2014 International Conference on Embedded Systems(ICES). IEEE, 2014:238-243.
- [4]Williams D P. Underwater target classification in synthetic aperture sonar imagery using deep convolutional neural networks[C]//2016 23rd international conference on pattern recognition(ICPR). IEEE, 2016:2497-2502.
- [5]Zhou T, Si J, Wang L, et al. Automatic detection of underwater small targets using forward-looking sonar images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60:1-12.
- [6]Wang Z, Zhang S, Huang W, et al. Sonar image target detection based on adaptive global feature enhancement network[J]. IEEE Sensors Journal, 2021, 22(2):1509-1530.
- [7]Tucker J D, Azimi-Sadjadi M R. Coherence-based underwater target detection from multiple disparate sonar platforms[J]. IEEE Journal of Oceanic Engineering, 2011, 36(1):37-51.
- [8]Kong W, Hong J, Jia M, et al. YOLOv3-DPFIN:A dual-path feature fusion neural network for robust real-time sonar target detection[J]. IEEE Sensors Journal, 2019, 20(7):3745-3756.
- [9]Zhao L, Li S. Object detection algorithm based on improved YOLOv3[J]. Electronics, 2020, 9(3):537.
- [10]Qiu M, Huang L, Tang B H. ASFF-YOLOv5:Multielement detection method for road traffic in UAV images based on multiscale feature fusion[J]. Remote Sensing, 2022, 14(14):3498.
- [11]El Jurdi R, Petitjean C, Honeine P, et al. CoordConv-unet:investigating CoordConv for organ segmentation[J]. IRBM, 2021, 42(6):415-423.
- [12]Lu J, Ma C, Li L, et al. A vehicle detection method for aerial image based on YOLO[J]. Journal of Computer and Communications, 2018, 6(11):98-107.
- [13]Sang J, Wu Z, Guo P, et al. An improved YOLOv2 for vehicle detection[J]. Sensors, 2018, 18(12):4272.Moore D, Rid T. Cryptopolitik and the Darknet[J]. Survival, 2016, 58(1):7-38.
- [14]Wang K, Liew J H, Zou Y, et al. Panet:Few-shot image semantic segmentation with prototype alignment[C]//proceedings of the IEEE/CVF international conference on computer vision. 2019:9197-9206.
- [15]Bochkovskiy A, Wang C Y, Liao H Y M. Yolov4:Optimal speed and accuracy of object detection[J]. arXiv preprint arXiv:2004.10934, 2020.
- [16]Zhang P, Li D. CBAM+ASFF-YOLOXs:An improved YOLOXs for guiding agronomic operation based on the identification of key growth stages of lettuce[J]. Computers and Electronics in Agriculture, 2022, 203:107491.
- [17]Yin A, Ren C, Yue W, et al. CDAU-Net:A Novel CoordConv-Integrated Deep Dual Cross Attention Mechanism for Enhanced Road Extraction in Remote Sensing Imagery[J]. Remote Sensing, 2023, 15(20):4914.
- [18]周彦，陈少昌，吴可，等. SCTD1. 0:声呐常见目标检测数据集[J].计算机科学，48(11A):334-339.
- [19]Chen Z, Tang J, Qiu H, et al. MGFGNet:an automatic underwater acoustic target recognition method based on the multi-gradient flow global feature enhancement network[J]. Frontiers in Marine Science, 2023.
- [20]Tang J, Chen Z, Xie H, et al. Unbalanced Underwater Sonar Image Target Detection Based on Weak Contrast Feature Enhancement[C]//2023 IEEE 7th Information Technology and Mechatronics Engineering Conference(ITOEC). IEEE, 2023, 7:367-373.