基于超像素的海面油膜分割应用  

Application of Sea Oil Film Segmentation Based on Superpixel

作　　者：靳卫卫 刘保占 赵建平 栗宝鹃 于新生[3] 邱照宇 JIN Wei-wei;LIU Bao-zhan;ZHAO Jian-ping;LI Bao-juan;YU Xin-sheng;QIU Zhao-yu(Safety&Environmental Protection Branch,CNOOC Energy Technology&Services Ltd.,Tianjin 300452,China;China Water Resources Beifang Investigation,Design&Research Co.Ltd.,Tianjin 300222,China;Ocean University of China College of Marine Geosciences,Qingdao Shandong 266100,China)

机构地区：[1]中海油能源发展股份有限公司安全环保分公司,天津300452 [2]中水北方勘测设计研究有限责任公司,天津300222 [3]中国海洋大学海洋地球科学学院,山东青岛266100

出　　处：《船海工程》2020年第2期52-54,共3页Ship & Ocean Engineering

基　　金：国家重点研发计划(2017YFC1405600);国家自然科学基金(41776182);中海石油环保服务有限公司项目(RLKJ-2014-01)。

摘　　要：为解决海面多噪声环境下油膜目标快速提取问题,兼顾分割效果和效率,提出基于区域特性的快速分割应用。采用超像素构建区域级语义图像,以像素区块特征形成新的相似度量单元,对超像素进行聚类分割,完成海面油膜的快速分割处理。结果显示,此方法可以有效地完成海面油膜目标的分割提取。基于超像素的分割技术能够兼顾图像区域特性和处理效率,具备抗噪能力优势,提高海面油膜分割算法精度。In order to solve the problem of rapid extraction of oil film targets in a sea-noisy environment,a fast segmentation application based on superpixels was proposed.A super-pixel was used to construct a regional-level semantic image,and a new similarity measurement unit was formed based on the characteristics of the pixel block.The super-pixel was clustered and segmented with the binary K-means algorithm to complete the rapid segmentation of the sea surface oil film.The results showed that this method can effectively complete the segmentation and extraction of sea surface oil film targets.The superpixel-based segmentation technology can take into account the image area characteristics and processing efficiency,and has the advantage of anti-noise capability,improves the accuracy of the sea surface oil film segmentation algorithm.

关 键 词：超像素 暗通道去雾 K均值聚类 油膜分割 

分 类 号：U698.7[交通运输工程—港口、海岸及近海工程]