基于改进流域算法的三维图像分割方法

3D Image Segmentation Based on Improved Watershed Algorithm

在石油地质研究中,分析由CT扫描真实岩心所得到的三维结构已成为研究岩心微观孔隙结构的重要手段。由于噪声因素等影响,三维重建所得的岩心颗粒目标往往存在粘连情况,严重影响了后续的分析。为此,借鉴流域算法的思想并加以改进,提出了一种快速且有效的三维图像分割算法。算法中采用广度优先搜索算法(BFS算法)进行区域增长,提高了运算速度;同时采用基于颗粒间重叠程度的区域合并方法,有效抑制了过分割现象。最后利用算法对岩心三维图像进行了分割。实验证明了该算法的正确性和实用性。

It＇ s a important method that analyzing the 3D microscopic pore structure which is scaned by CT in petroleum geology research. Because of noise, there exist adhesions between the 3 D core particles, which have seri- ous impact on calculation. A fast and effective 3D image segmentation algorithm is proposed. The algorithm apply breadth-first search algorithm to regional growth in order to improve the computing speed. At the same time, region merging method based on the degree of overlap between the particles in order to inhibit the over-segmentation is proposed. At last, the core 3D image is splited using the algorithm which proposed and compare the result with the skeletonization segmentation algorithm＇ s and the maximal ball algorithm＇ s. The experiments prove the correctness and practicality of the algorithm.