基于双路卷积神经网络的高分辨率遥感影像浒苔检测方法
Ulva polifera Detection from High Resolution Remote Sensing Images Based on Dual-Path Convolutional Neural Networks
王艳丽 1董志鹏 2王密3
作者信息
- 1. 山东科技大学测绘与空间信息学院,山东 青岛,266590
- 2. 自然资源部第一海洋研究所,山东 青岛,266061
- 3. 武汉大学测绘遥感信息工程国家重点实验室,湖北 武汉,430079
- 折叠
摘要
浒苔绿潮是一种危害巨大的海洋生态灾害.如何快速准确地检测出浒苔,对其及时治理和促进海洋产业健康发展具有重要意义.针对高分辨率遥感影像浒苔检测中浒苔区域边界难以精确确定的问题,提出一种基于双路卷积神经网络的高分辨率遥感影像浒苔检测方法.首先,基于高分辨率遥感影像中浒苔分布特性,设计了一种双路卷积神经网络语义分割架构,用于提取影像中浒苔区域和边界等属性特征;然后,提出一种基于浒苔边界辅助优化浒苔区域检测结果策略,对初始浒苔检测结果优化处理获得精确浒苔区域检测结果;最后,进行定性对比实验和定量评价.结果表明,所提方法对高分辨率遥感影像浒苔检测结果的F1分数、交并比和整体分类精度分别为88.25％、78.97％、98.99％,能够实现对不同类型的浒苔精确检测,获得良好的高分辨率遥感影像浒苔检测结果.
Abstract
Objectives:The green tide formed by Ulva prolifera(U.prolifera)is a harmful marine ecologi-cal disaster.The rapid and accurate detection is of great significance for timely management of U.prolifera and the healthy development of the marine industry.Methods:Because the boundary of U.prolifera area is difficult to be determined accurately in high resolution remote sensing images(HSRIs),an U.prolifera de-tection method for HSRIs based on dual-path convolutional neural networks(CNN)is proposed in this pa-per.First,a dual-path CNN semantic segmentation framework is designed based on the distribution charac-teristics of U.prolifera in HSRIs.The area and boundary of U.prolifera in HSRIs can be extracted simulta-neously using the proposed framework.Then,the strategy for optimizing the initial U.prolifera area de-tection results based on U.prolifera boundary is proposed to improve the detection accuracy.Results:The experimental results show that the proposed method can extract U.prolifera accurately,with F1-score of 88.25％,intersection-over-union of 78.97％and over accuracy of 98.99％,which is better than other U.prolifera detection algorithms.Conclusions:The proposed method can obtain good results for the detection of different types of U.prolifera in HSRIs.
关键词
浒苔/高分辨率遥感影像/卷积神经网络/信息提取/多特征融合Key words
Ulva prolifera/high resolution remote sensing image/convolutional neural network(CNN)/information extraction/multi-feature fusion引用本文复制引用
出版年
2024
NETL
NSTL
万方数据