光学与光电技术, 2010, 8 (6): 6, 网络出版: 2011-02-15
基于分步区域增长算法的图像自动分割方法
Methods of Image Automatic Segmentation Based on Step Region-Growth
分步区域增长 图像自动分割 陆海分离 海岸线提取 step region growth image automatic segmentation land-sea separation coastline extraction
摘要
利用遥感图像进行海上目标检测,关键的问题之一就是陆海分离。如果采用传统的阈值分割方法,在陆地一些灰度较低的区域容易造成误分割。分裂合并算法可以解决误分割的问题,但是这种算法对于海面上的船只、珊瑚礁等干扰因素会产生一些空隙及孤立区域,所以要采用一些数学形态学的方法去干扰。而数学形态学的方法会损失图像的边缘信息量。为此提出了分步区域增长算法进行图像自动分割,通过海面陆地两次区域增长,成功地提取出了主体陆地区域。实验结果表明,该方法可以很好地分离出图像上的主体陆地区域,并有较快的处理速度。
Abstract
In the maritime target detection using remote image, one of the key problems is land-sea separation.Thetraditional method is threshold segmentation, but in the low gray scale areas of land it is easy to cause mis-segmentation. Divid-unit method can solve this problem, whereas if there are some interference on the sea, such as shipand coral reef, some isolated regions will be formed. So step region growth method is adopted for image segmentation.This method processes region growth twice to extract the main land region successfully. The experimental results showthat the method has fast process speed.
侯旺, 杨夏. 基于分步区域增长算法的图像自动分割方法[J]. 光学与光电技术, 2010, 8(6): 6. HOU Wang, YANG Xia. Methods of Image Automatic Segmentation Based on Step Region-Growth[J]. OPTICS & OPTOELECTRONIC TECHNOLOGY, 2010, 8(6): 6.