基于数学形态学的灰度图像连接物体分割方法

由于噪声的存在以及连接物体的特点,传统的标记分水岭算法对包含连接物体的灰度图像很难取得满意的分割结果;特别是在背景并不连通的情况下,误分割更为常见;在标记分水岭算法的基础上,提出了一种连接物体分割方法;将属于鲁棒统计的Hough变换用于提取物体标记扩展了标记分水岭算法的应用范围;针对在分割连接物体时,由于背景并非连通,因此允许背景被分别标记,并通过一个后续滤波步骤用以剔除分割后图像中的背景部分,从而得到精确的分割图像;试验证明该算法运算速度快,鲁棒性好,具有广泛的应用价值.

Grayscale Touching Objects Segmentation by Mathematical Morphological Methods

Due to image noise and the characteristics of images containing touching objects, traditional marker-driven watershed algorithm can not get good segmentation results, especially when the background is not connected. A touching object segmentation method based on mathematical morphology is proposed. By using Hough transform to extract inter markers expands the usability of marker-driven watershed algorithm. In images which contain touching objects, the background is not connected. So it is necessary to allow background to be separately marked, and a post-processing step is used to filter the background in the segmented image. It is proved that the algorithm is not only rapid and robust to parameters change, but also applicable to a wide range of applications.