基于卷积神经网络的光电导航图像超分辨率方法  

Photoelectric Navigation Image Super-Resolution MethodBased on Convolutional Neural Network

作　　者：惠心雨 汪辉 白俊强[1] 郭彬 刘成茂 HUI Xinyu;WANG Hui;BAI Junqiang;GUO Bin;LIU Chengmao

机构地区：[1]西北工业大学,西安710072 [2]西安索格亚航空科技有限公司,西安710065

出　　处：《现代导航》2020年第6期421-424,441,共5页Modern Navigation

摘　　要：本文设计了一种基于卷积神经网络的光电导航图像超分辨率系统,为飞机飞行或着陆、卫星图像提供可靠的导航信息。超清化的可见光图像能够为地面监控人员、飞行员或无人机提供可靠的图像信息。超清化方法可以降低导航系统对硬件设备、计算资源的要求,适用于普通民航客机与搭载低精度摄像头的小型无人机。本文以卷积神经网络作为框架,搭建接近实时计算的图像超分辨率模型。In this paper,a photoelectric navigation image super-resolution system based on convolutional neural network is designed to provide reliable navigation information for aircraft flight or landing and satellite images.Super definition visible images can provide reliable information for ground monitors,pilots or drones.The super definition method can reduce the requirements of navigation system on hardware equipment and computing resources,and is suitable for ordinary civil aircraft and small UAV with low precision cameras.In this paper,the convolutional neural network is used as a framework to build an image super-resolution model for near real-time computing.

关 键 词：机器学习 卷积神经网络 卫星图像 超清化 导航 

分 类 号：TN753[电子电信—电路与系统]