【作者】 洪志令； 姜青山； 董槐林； 王声瑞；

【机构】 厦门大学计算机科学系； 厦门大学软件学院；

【摘要】 <正>1引言常见的基于区域的图像分割方法有阈值分割,区域生长与分裂合并和聚类分割法。阈值分割法是一种简单有效的图像分割方法,最具代表性的阈值分割方法有直方图阈值、类间方差阈值、最大熵值、模糊聚类分割法等。更多还原

【Abstract】 In this paper,we propose a new multi-threshold image segmentation method based on two-dimensional histogram.The new method is also applied to high-resolution SAR imagery to detect ship.In this method,we treat histogram as relief map which is made of catchment basins and watersheds.By virtue of the algorithm of immersion simulations,we complete the multi-threshold image segmentation.Essentially,it’s a data cluster process which based on a hypothesis that the data are drawn from one of several probabilistic distributions.In our experiment,we compare our method with several existing popular methods.The experiment show that our method is not only direct, simple and less computing,but also can get the threshold exactly,it more fit for image segmentation.The application in SAR ship detection shows its validity.更多还原