基于MobileNetV2的水库沉底目标声呐图像分类方法  

A Classification Method of Sonar Images for Reservoir Sinking Target Based on MobileNetV2

作　　者：张美燕 许晓雯 傅淑丹 ZHANG Meiyan;XU Xiaowen;FU Shudan(Electrical Engineering Department,Zhejiang University of Water Resources and Electric Power,Hangzhou 310018,China)

机构地区：[1]浙江水利水电学院电气工程学院,浙江杭州310018

出　　处：《浙江水利水电学院学报》2024年第4期86-90,共5页Journal of Zhejiang University of Water Resources and Electric Power

基　　金：浙江省自然科学基金资助项目(LZJWY22E090001);校大学生创新创业训练计划项目(S202311481073)。

摘　　要：针对水库底部探测需求,提出一种基于MobileNetV2的水库沉底目标声呐图像分类方法,解决了声呐目标分类过程中受严重噪声和低信噪比干扰造成分类效果不佳的问题。6种典型水库底部目标声呐图像实际测试结果表明,该算法与VGG(Visual geometry group)、ResNet(Residual network)和Efficient Net等图形分类模型相比,分类准确率和召回率等指标都有所提高,计算复杂度更低。In response to the demand for reservoir bottom detection,a sonar image classification method for reservoir bottom targets is proposed based on MobileNetV2,which solves the problem of poor classification performance caused by severe noise and low signal-to-noise ratio interference in the sonar target classification process.The actual test results of 6 typical reservoir bottom target sonar images show that compared with VGG,ResNet,Efficient Net and other graphical classification models,the algorithm proposed in this paper has improved classification accuracy and recall rate with lower computational complexity.

关 键 词：MobileNetV2 声呐图像 水库沉底目标 目标分类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]