基于复杂背景的图像分割算法

Algorithm of Image Segment Based on Complicated Background

针对圆柱形钢棒在线计数特定的复杂背景提出了基于双阈值结合数学形态学运算的分割算法。首先分别根据2个阈值进行分割得到2幅二值图像，然后根据目标是圆形以及其大致位置等先验知识分别对2幅图进行腐蚀，开启，闭合等数学形态学运算，最后将高阈值分割得到的图像中的边界信息叠加到低阈值分割得到的图像中。实验结果表明，本方法作为本系统的图像预处理是有效的。

This paper introduces a new method based on double-threshold segmenting combined with mathematical morphology for the column form steel product online counting system, which always has a complicated background. First, use two thresholds to get two binary images; then, conduct erosion, opening and closing in these two images with some specially shaped factors that have been known about the shape and location of the objects in the image. Lastly, combine the edge information detected from the image which has been computed with a low threshold with the one detected from another image. From the result of experimentation, it can be found that the new method works well as the pre-processing part in the counting system.