摘要
介绍了一种基于目标灰度门限和目标之间灰度距离门限的区域自动阈值检测法,用该方法检测出图像区域的阈值进行目标初分割,然后结合形态学中的开启和闭合方法对初分割后的二值图像进行双滤波,再用一种新的区域边缘跟踪标注法对其进行跟踪和标注,找出每个目标的包络矩形坐标,用其对原图进行区域定位,从而可以提取出原图中包含目标的小区域.
First, this paper presents a new method for automatically detecting threshold for image segmentation. Then image is segmented based on the threshold that is obtained by applying this new method. Afterward image is filtered using opening and closing algorithm, which are derived from mathematical morphology. Whereafter, the objects are labeled by a new edge tracing method and the coordinates of the least rectangle enveloping every object are detected. Finally these coordinates are used to get the 'local area' that envelops the object in the source image. The feasibility of the approach is proved experimentally.
出处
《装甲兵工程学院学报》
2004年第4期55-57,65,共4页
Journal of Academy of Armored Force Engineering
关键词
阈值分割
图像开运算
图像闭运算
图像标记
图像跟踪
<Keyword>threshold segmentation
opening operation
closing operation
image label
image track