摘要
针对传统飞机目标检测算法和现有机器学习检测算法存在的问题,提出了遥感影像中的一种有效飞机检测概念;在认知模型下使用基于深度学习的全卷积检测和分割网络,设计了一种有效飞机目标检测系统并对其进行了仿真;构建了一种检测认知模型,并设计了各模块的功能。实验结果证明了该系统的有效性,该系统为开展目标智能检测提供了一种全新的思路和方法。
In view of the problems in the traditional aircraft detection algorithms and the existing machine learning detection algorithms,one concept of valid aircraft detection is proposed for remote sensing images.With the full convolution detection and segmentation network based on the deep learning in the cognitive model,one valid aircraft detection system is designed and simulated.A cognitive model for detection is constructed,and the function of each module is designed.The experimental results certify the effectiveness of this system,and this system provides a new thinking way and method for the development of intelligent detection of multiple objectives.
作者
侯宇青阳
全吉成
魏湧明
Hou Yuqlngyang;Quan Jicheng;Wei Yongming(Laboratory of Digital Earth Science, Aviation University of Air Force, Changchun, gilin 130000, China)
出处
《光学学报》
EI
CAS
CSCD
北大核心
2018年第1期134-143,共10页
Acta Optica Sinica
关键词
成像系统
有效飞机检测
深度学习
认知模型
遥感影像
图像分割
imaging systems
valid aircraft detection
deep learning
cognitive model
remote sensing images
image segmentation
