基于Radon变换和梯度跟踪的海岸线提取算法  

Shoreline Detection Method in Video Images Based on Radon Transform and Gradient Tracking Algorithm

作　　者：张锁平[1,2] 张春田[1] 

机构地区：[1]天津大学电子信息工程学院,天津300072 [2]国家海洋技术中心,天津300112

出　　处：《光电工程》2009年第3期57-62,共6页Opto-Electronic Engineering

基　　金：国家人事部留学回国人员科技活动项目择优资助课题;国家海洋公益项目(200805018)

摘　　要：针对视频图像中自动提取海岸线的问题,提出了一种基于Radon变换和彩色图像梯度跟踪模型的海岸线提取算法。根据海水和陆地的彩色图像信息差异,在计算彩色梯度向量的基础上,利用Radon变换检测线状奇异性的特点,设计剖面分析方法,预测水陆交界处大致方向和动态范围,以达到缩小检测区域的目的;最后采用一种改进的最大梯度跟踪模型对海岸线进行搜索,以提取最终的海岸线精确定位参数。实验结果验证了该算法对海岸线检测的有效性和较强的鲁棒性。To extract shoreline from video monitoring images, a method is proposed by applying Radon transform and gradient tracing algorithm. Based on the different color properties of land and water image, the gradient vector is calculated to present edges in color space efficiently. Then, with Radon transform and its basic ability of linear singularity detection, a profile analysis technique is used to further select the corresponding peak area. Meanwhile, the approximate direction and area of waterline is obtained in the Radon domain. In addition, the image region including interface of shore and sea boundaries is located by reconstructing image using inverse Radon transform. Finally, the shoreline is determined by an improved searching strategy to explore the maximum gradient path. Experimental results indicate that the proposed approach is effective in finding the shorelines and the algorithm has robustness with varies lightening conditions.

关 键 词：海岸线 彩色空间 边缘检测 RADON变换 梯度跟踪 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] P71[自动化与计算机技术—计算机科学与技术]