基于深度卷积神经网络的可见光舰船目标识别系统

Ship Recognition From Infrared Images Based on Deep Convolutional Neural Network

论文针对日益复杂的海面战场环境对舰船目标识别更高识别率、实时性、智能化的需求,进行了基于深度卷积神经网络的可见光舰船目标识别系统的方案设计。可见光舰船目标识别系统由舰船图像采集预处理平台、舰船图像离线训练平台和舰船目标在线识别平台组成,可进行海面舰船、民船的可见光/红外图像采集、处理和存储,可进行基于深度卷积神经网络的舰船目标特征训练和识别,具有较高的目标识别准确率和实时性。

Due to the complex sea battlefield,ship recognition should be more high-accuracy,real-tme and intelligent,a novel ship recognition system from infrared images is proposod based on deep convolutional neural network.The system is consisted of ship image acquisition and preprocessing platform,offline training platform and online target recognition platform.The system can acquire,process and store visible or infrared images of warship and civil ship,and feature training and recognition of ship tar⁃gets are carried out.The proposed system is real-time with high recognition accuracy.