基于分数阶微分和形态学多级合成的岩石节理裂隙图像分割  

Rock joint image segmentation based on fractional differential and multi-grade combination in mathematical morphology

作　　者：王卫星[1] 谭嘉玉[1] 

机构地区：[1]重庆邮电大学计算机科学与技术学院,重庆400065

出　　处：《计算机应用》2010年第4期929-931,942,共4页journal of Computer Applications

基　　金：国家自然科学基金资助项目(60873186/F020509)

摘　　要：岩石节理裂隙形状复杂且无规则,图像中带有大量噪声,利用传统的图像分割方法很难达到很好的分割效果。提出了一种基于分数阶微分和数学形态学多级合成的边缘检测方法。首先对岩石裂隙图像进行噪声滤除、图像分割、空腔填充、短枝去除等操作,然后使用分数阶微分的方法进行预处理,最后采用改进形态学多级合成方法得到结果。实验结果表明,该方法与传统算子相比,对岩石节理裂隙图像具有较好的边缘检测能力和抗噪性。Rock joint network is very complex and there is much noise in a rock image.The traditional image segmentation methods cannot obtain satisfactory results.A new edge detection algorithm was proposed based on fractional differential and multi-grade synthesis of mathematical morphology.It needs to pretreat the rock fracture image first by image processing operation such as noise filtering,image segmentation,cavity filling,spur removal,etc.Then fractional differential algorithm and multi-grade synthesis of mathematical morphology were used to get the results.The experimental results,compared with the traditional morphological methods,show that the studied algorithm maintains good detected edges of rock joint images and can increase the denoising power as well.

关 键 词：图像分割 分数阶微分 数学形态学 多级合成 岩石节理裂隙 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]