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.
A Classification Method of Sonar Images for Reservoir Sinking Target Based on MobileNetV2
In response to the demand for reservoir bottom detection,a sonar image classification method for reser-voir bottom targets is proposed based on MobileNetV2,which solves the problem of poor classification perform-ance 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 classifi-cation accuracy and recall rate with lower computational complexity.
NETL
NSTL
万方数据
