基于标记的数学形态学滤波分水岭算法  

Mathematical-morphological Watershed Algorithm Based on Marking

作　　者：蔡子文[1] 费向东[1] 

机构地区：[1]四川大学计算机学院视觉合成图形图像技术重点学科实验室,四川成都610064

出　　处：《计算机技术与发展》2013年第3期38-40,44,共4页Computer Technology and Development

基　　金：国家"973"重点基础研究发展计划项目(2009CB320803)

摘　　要：经典的分水岭算法存在过分割问题,文中针对图像分割提出了一种基于预处理的改进算法。目的是为了抑制分割过程中的过分割现象,缓解过分割问题。首先对图像进行应用数学形态学去噪,滤波。再求取梯度图像,然后根据梯度图像局部极小值的综合信息自适应地提取内部标记,再进行距离分水岭变换提取外部标记。并以提取的标记为依据,对梯度图像进行修正,最后对修正后的梯度图像实施分水岭算法。实验结果表明:与传统分水岭算法相比,本算法能较好地缓解过分割问题。The oversegmentation is a problem of classical watershed algorithm,it presents an improved watershed algorithm based on pre processing for image segmentation. The purpose is to suppress oversegmentation phenomena in the process of segmentation,alleviate the oversegmentation problem. Firstly, the morphological is used to denoise and filter. Then calculate the gradient image, subsequently, adaptively extract internal marker according to the integrated information of gradient image local minimum value, then carry on the dis tance watershed transformation to extract external marker. Based on the marker, correct the gradient image, finally, the watershed algo rithm is used to gradient image corrected. The result shows that compared with traditional watershed algorithm, this method can effectively remit the oversegmentation problem.

关 键 词：数学形态学 分水岭算法 图像分割 过分割 梯度图像 标记提取 

分 类 号：TP31[自动化与计算机技术—计算机软件与理论]