摘要
在海洋环境污染中,溢油在发生频率、损害程度、长期负面效应等方面均居首位。为了减少溢油事故带来的损失,在有效的时间内采用先进的技术实时监测海上溢油是应急处置和快速恢复海洋生态环境的重要保障。航海雷达溢油监测技术具有全天时、全天候随船作业的优势,可以有效辅助溢油事故的应急处置。在近10年来,这项技术逐渐受到国内外学者的关注。本文从原始数据预处理、溢油有效监测区提取、油膜分割等方面对航海雷达溢油监测技术的研究进行综述与分析。采用图像纹理特征、机器学习分类与自适应阈值分割相结合的溢油识别方法是目前航海雷达溢油监测的主流技术。随着样本数量的不断积累,应用深度学习方法进行航海雷达图像海上溢油的实时语义分割是未来发展的主要方向。
Oil spill ranks first in terms of occurrence frequency,pollution degree and long-term negative effects in marine environmental pollution.To minimize the losses after the oil spill accident,advanced technology for monitoring the oil spill in real-time is an important guarantee for emergency treatment and rapid recovery of marine ecological environment.The marine radar oil spill monitoring technology has the advantage of assisting emergency response all-day and all-weather.It has gradually attracted attention in the past decade.The marine radar oil spill monitoring technology is summarized and analyzed from the aspects of data preprocessing,effective region extraction,oil film segmentation.The oil spill identification method based on image texture features,machine learning and adaptive threshold is the mainstream in marine radar oil spill monitoring technology.With the continuous accumulation of the image samples,marine radar oil spill semantic segmentation using deep learning is the main direction in the future.
作者
徐进
李博
陈蓉
马龙
赵志强
李希栋
XU Jin;LI Bo;CHEN Rong;MA Long;ZHAO Zhiqiang;LI Xidong(Naval Architecture and Shipping College,Guangdong Ocean University,Zhanjiang 524088,China;Shenzhen Institute of Guangdong Ocean University,Shenzhen 518120,China)
出处
《海洋环境科学》
CAS
CSCD
北大核心
2023年第5期805-812,共8页
Marine Environmental Science
基金
广东省自然科学基金项目(2022A1515011603)
广东省普通高校重点领域专项(2022ZDZX3005)
深圳市基础研究专项—自然科学基金项目(JCYJ20220530162200001)
广东海洋大学科研启动经费资助项目(060302132009)。
关键词
航海雷达
溢油
机器学习
纹理特征
marine radar
oil spill
machine learning
texture feature
