CHEN Chunshan,LI Yingxiang,ZHONG Jiandan.A Flood Detection Method based on Image Semantic Segmentation[J].Journal of Chengdu University of Information Technology,2025,40(02):151-156.[doi:10.16836/j.cnki.jcuit.2025.02.005]
一种基于图像语义分割的洪水检测方法
- Title:
- A Flood Detection Method based on Image Semantic Segmentation
- 文章编号:
- 2096-1618(2025)02-0151-06
- Keywords:
- flood detection; region growth; maximum entropy; semantic segmentation
- 分类号:
- TP391.41
- 文献标志码:
- A
- 摘要:
- 为了能以图像的方式对洪水发生情况进行有效检测,提出一种基于图像语义分割的洪水检测方法。首先对洪水图像的Lab空间L分量和RGB空间R分量进行最大熵分割,将分割结果进行逻辑与运算,然后利用形态学腐蚀去除水纹噪点欠分割和河岸边缘过分割的影响得到预处理图像; 在此基础上提出改进区域生长的洪水分割算法,根据洪水区域的分布特点得到初始生长区域,利用图像最大熵设置区域生长的自适应阈值,进行区域生长得到洪水区域分割图像。此分割算法的准确率和平衡F分数分别为97.05%、97.02%,平均交并比为94.2%,能有效完成洪水图像分割。同时,设计了一种洪水溢流器,能根据算法分割出来的洪水区域,在未发生洪水的基础上计算洪水溢流比,将洪水划为警示、警告、危险3个等级,完成洪水检测。
- Abstract:
- To effectively detect the occurrence of floods through images, this paper proposed a flood detection method based on image semantic segmentation. Firstly, the method used the maximum entropy segmentation on the L component of Lab space and R component of RGB space to segment the flood image, and multiplied the segmentation results. Then, it used morphological erosion to remove the effects of under-segmentation of water wave noise and over-segmentation of riverbank edges, after the above operation, it got a preprocessed image. On this basis, this paper proposed an improved region growth flood segmentation algorithm, according to the distribution characteristics of the flood area, it extracted the initial growth area, so that it obtained the segmentation image of the flood area by using an adaptive threshold which got from the maximum entropy. The accuracy and F-score of this segmentation algorithm reached 97.05%、97.02%,and mIoU reached 94.2%,it can effectively complete flood image segmentation. This paper also proposed a method of flood overflow calculation, which can calculate the flood overflow ratio based on the flood segmentation algorithm, and classify the flood into caution, warning, and danger levels, it completed the flood detection.
参考文献/References:
[1] 张晴丹.专家警告“汛期雨季不进山”[N].中国科学报,2022-08-17.
[2] Basnyat B,Roy N,Gangopadhyay A.Flood Detection using Semantic Segmentation and Multimodal Data Fusion[C].2021 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events(PerCom Workshops).IEEE,2021.
[3] Co-Operationdevelopment O.Financial Management of Flood Risks[J].Water Intelligence Online,2016,15:1-138.
[4] CRED.2022 Disasters in numbers[R].Brussels:CRED,2023.
[5] Shen X,Wang D,Mao K,et al.Inundation Extent Mapping by Synthetic Aperture Radar:A Review[J].Remote Sensing,2019,11(7):879.
[6] 苗添,曾虹程,王贺,等.基于迭代阈值分割的星载SAR洪水区域快速提取[J].系统工程与电子技术,2022,44(9):2760-2768.
[7] Filonenko A,Wahyono,DC Hernández,et al.Real-time flood detection for video surveillance[C].Conference of the IEEE Industrial Electronics Society.IEEE,2016.
[8] Lo S W,Wu J H,Lin F P,et al.Cyber Surveillance for Flood Disasters[J].Sensors,2015,15(2):2369-2387.
[9] Menon K P,Kala L.Video surveillance system for realtime flood detection and mobile app for flood alert[C]. International Conference on Computing Methodologies and Communication,2017.
[10] Wan A A,Pebrianti D,Ronny,et al.Image Processing-Based Flood Detection[C].Nationaltechnical seminar on underwater system technology.Magister of Computer Science,Universitas Budi Luhur,Jakarta 12260,Indonesia; Magister of Computer Science,Universitas Budi Luhur,Jakarta12260,Indonesia,Faculty of Electrical andElectronics Engineering,Universiti Malaysia Pahang,Pekan,Malaysia; Fac,2019.
[11] Pereira J,Monteiro J,Silva J,et al.Assessing flood severity from crowdsourced social media photos with deep neural networks[J].Multimedia Tools and Applications,2020,79(35/36).
[12] Huang J,Kang J,Wang H,et al.A Novel Approach to Measuring Urban Waterlogging Depth fromImages Based on Mask Region-Based Convolutional Neural Network[J].Sustainability,2020,12.
[13] 易三莉,张桂芳,贺建峰,等.基于最大类间方差的最大熵图像分割[J].计算机工程与科学,2018,40(10):1874-1881.
[14] 杨超,刘本永.基于Lab颜色空间纹理特征的图像前后景分离[J].激光与光电子学进展,2019,56(12):59-64.
[15] 吴骅跃,段里仁.基于RGB熵和改进区域生长的非结构化道路识别方法[J].吉林大学学报:(工学版),2019(3):727-735.
[16] Satoshi,Suzuki.Topological structural analysis of digitized binary images by border following[J].Computer Vision Graphics & Image Processing,1985,30(1):32-46.
[17] 于晓,吕欣欣,高强,等.基于最大熵生长检测器的模糊红外图像分割算法[J].激光杂志,2019,40(3):6.
[18] 黄鹤,茹锋,王会峰,等.基于单峰偏移最大熵阈值分割雾天道路能见度检测方法[P].CN111275698A,2020.
[19] Mijic A.Deep Learning Semantic Segmentation for Water Level Estimation Using Surveillance Camera[J].Applied Sciences,2021,11.
[20] Barz B,Kai S,M Münch,et al.Enhancing Flood Impact Analysis using Interactive Retrieval of Social Media Images[J].arXiv e-prints,2019.
[21] Liang Y,Li X,Tsai B,et al.V-FloodNet:A video segmentation system for urban flood detection and quantification[J].Environmental Modelling & Software,2023,160:105586.
[22] Liang Y,Jafari N,Luo X,et al.WaterNet:An adaptive matching pipeline for segmenting water with volatile appearance[J].Computational Visual Media,2020(1):14.
备注/Memo
收稿日期:2023-10-18
基金项目:四川省科技计划资助项目(2023YFS0428、2023YFG0160)