基于模糊距离变换的岩心图像颗粒分割算法

The grain image segment algorithm based on the fuzzy distance transformation

岩心CT图像中相邻颗粒之间存在相互粘连,粒度参数分析等需要对粘连颗粒进行分割。结合岩心粘连颗粒的特性,提出一种基于模糊距离变换的改进颗粒分割算法。首先,对预处理后的岩心图像进行模糊距离变换并提取出距离信息的灰度图像,然后利用形态学膨胀重构方法提取标记作为后续分水岭算法的种子点,根据种子点采用一种基于测地重建的改进分水岭算法得到相邻种子点之间的分割线,最后完成粘连颗粒的分割。实验结果表明本文算法可以提高粘连颗粒分割的准确度并减轻分水岭算法的过分割现象。

There is a phenomenon in CT image of rock that the core grains often adhere each other closely. Adhesion segment should be done for the parameter analysis of grains. According to the features of adhesion grains, this paper presents an improved grain segment algorithm based on fuzzy distance transformation. Firstly, by using fuzzy distance transformation to rock image, the distance information in gray image will be obtained. Then a morphological expansion reconstruction algorithm is used to abstract the central seeds which are important in watershed algo- rithm. The dividing line among adjacent seeds can be figureed out with the geodesic reconstruction algorithm based on the central seeds. The experimental results indicate that this algorithm can accurately and effectively split the adhesion among the grains and reduce the over-segmen- tion of the watershed algorithm.