光学与光电技术, 2018, 16 (2): 20, 网络出版: 2018-06-01  

基于深度卷积神经网络的海战场目标协同识别方法

Collaborative Recognition Method for Sea Battlefield Target Based on Deep Convolutional Neural Networks
作者单位
华中光电技术研究所—武汉光电国家实验室, 湖北 武汉 430223
摘要
海战场的目标识别是敌我判断、精确跟踪以及准确打击的前提,在现代海战中起着至关重要的作用。针对舰船编队场景下的海战场典型目标的识别问题,提出了一种基于卷积神经网络的协同识别方法,该方法通过优化VGG-NET的层数、节点数得到一组易于训练的精简网络,构建了一种多输入/单输出的舰船协同识别架构,提出一种基于D-S证据理论的加权和方法用于单输出的决策级融合。仿真实验表明,基于单平台的方法在烟雾环境下仅获得47.22%的识别率,而使用协同识别方法能够获得55.83%的识别率。该方法具有良好的性能,能够在复杂海战场环境下有效地识别目标。
Abstract
Target recognition of sea battlefield is the prerequisite for the precise tracking and the judgment of the enemy, which is significant in modern naval warfare. For identifying the typical targets of sea battlefields under ship formation, a method of collaborative recognition based on convolutional neural network is proposed. Based on the optimized convolutional neural network and weighted D-S evidence theory, a multi-input/single-output ship co-recognition structure is constructed. The simulation results show that the single-platform method only obtains the accurary of 47. 22% in the smoke environment, while the accuracy of 55. 83% can be obtained by using the collaborative recognition method.

郑光迪, 潘明波, 刘巍, 吴学铜. 基于深度卷积神经网络的海战场目标协同识别方法[J]. 光学与光电技术, 2018, 16(2): 20. ZHENG Guang-di, PAN Ming-bo, LIU Wei, WU Xue-tong. Collaborative Recognition Method for Sea Battlefield Target Based on Deep Convolutional Neural Networks[J]. OPTICS & OPTOELECTRONIC TECHNOLOGY, 2018, 16(2): 20.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!