基于区域合并的灰度图像目标分离

Based on Regional Integration to Extract the Objects from Gray-level Images

针对背景模糊的灰度图像中目标物体往往难以分离出来的问题,文章采用基于区域生长的形态学分水岭算法获得模糊图像中的物体边界,然后针对分水岭变换后所获得的子区域的灰度值参数及所处位置特性,将每个子区域类比成3D地图表面的一块平原,采用类似山地斜坡平滑的思想将子区域逐渐合并,最终将目标从灰度图像中分离出来。实验表明该算法能准确的将较为模糊的灰度图像中的目标和背景分离。

For the issue of the background of blurred gray-level image the target object is often difficult to extract,this paper proposes a method to extract the objects of interest from an image and to distinguish the foreground from the background,starting from a partition of the image obtained by means of watershed transformation.Then for the watershed transform obtained after the sub-region of gray value parameters and location characteristics similar to smooth the slope ideas will gradually merge sub-region and,ultimately,in line with the targets set binary image.Experiments show that the algorithm will be more accurate fuzzy gray-level image of the object and background separation.