首页|协同U-Net与隐马尔可夫模型的遥感影像云识别

协同U-Net与隐马尔可夫模型的遥感影像云识别

扫码查看
云雾覆盖是影响光学遥感影像林业监测利用率的重要因素之一.文章针对传统云识别方法对噪声敏感,深度学习方法对云边缘识别精度不高的问题,提出了一种协同U-Net与隐马尔可夫模型的遥感影像云识别方法.首先,基于U-Net网络结构对云进行初步识别,改善传统方法对噪声的敏感性;其次,利用隐马尔可夫模型进行后端处理,优化云识别的边缘轮廓.实验结果表明,协同U-Net与隐马尔可夫模型的遥感影像云识别方法的精度相较于传统方法提升了5%,同时较好地保留了云的边缘轮廓.
Remote Sensing Image Cloud Identification of Cooperation with U-Net and Hidden Markov Model
Cloud and fog cover is one of the important factors affecting the utilization rate of optical remote sensing image forestry monitoring.In view of the problems that the traditional cloud recognition method is sensitive to noise and the Deep Learning method is not accurate in cloud edge recognition,a remote sensing image cloud recognition method of cooperation with U-Net and Hidden Markov Model is proposed in this paper.Firstly,the cloud is preliminarily identified based on the U-Net network structure to improve the sensitivity of the traditional method to noise.Secondly,it uses the Hidden Markov Model for back-end processing to optimize the edge contour of cloud recognition.The experimental results show that the accuracy of the remote sensing image cloud recognition method of cooperation with U-Net and Hidden Markov Model is improved by 5%compared with the traditional method.At the same time,the edge contour of the cloud is better preserved.

remote sensing imagecloud identificationU-Net Neural NetworkHidden Markov Model

程志强、崔成玲

展开 >

北京吉威空间信息股份有限公司,北京 100043

遥感影像 云识别 U-Net神经网络 隐马尔可夫模型

2024

现代信息科技
广东省电子学会

现代信息科技

ISSN:2096-4706
年,卷(期):2024.8(24)
  • 1