基于深层神经网络的军事目标图像分类技术

摘要

Aimed at demand of self discovery target function on unmanned platform, combined with deep learning method, an automatic military target classification method is proposed and simulated. Through experiment and test across six kinds of military targets, the proposed method has the advantages of accurate and fast classification and recognition, which can basically meet the demand of self discovery target function.%针对无人平台上自主发现目标功能的需求，结合深度学习方法，提出了一种军事目标图像分类技术思路，并且进行了模拟仿真。通过对六大类军事目标实验与测试，所提技术在特定训练库和特定像素图像的仿真验证中，识别精度和检测速度有所提高。