基于K-means岩石铸体图像分割及孔隙度的计算

为了准确获取储层的孔隙度进行地层解释并建立地质模型，设计了基于k-means的岩石铸体图像分割及孔隙度的计算方法。本设计基于k-means聚类算法对彩色铸体薄片进行有效分割，并且在分割基础上结合形态学相关知识对图像进行更加精确识别，通过计算机判读二值图像中的孔隙面积与总图像面积比值得到孔隙度值。实验结果表明该方法可以取得好的聚类分割效果，并且使用其他检测方法和计算机判读2种方法求得的孔隙度基本一致，数值较吻合。

K-means-based rock casting image segmentation and calculation of porosity

In order to accurately obtain the porosity of the deposited layer to carry out stratigraphic interpretation and set up a geological model,a calculation method is designed,in which rock casting body image is segmented so that the porosity is calculated based on K-means clustering method.The color casting body chips are effectively segmented with K-means clustering method and the image target can be recognized precisely in combination with the morphological knowledge on the basis of the segmentation.Thus the porosity is obtained by calculating the ratio of the porosity′s area to total image area.Experimental results show that this method not only has good clustering segmentation effect but also can obtain the values basically coincident with the ones from other detection methods.