基于图像分割的成像测井资料目标拾取与计算

为了从FMI资料中提取定量信息,一个很重要的步骤就是要对FMI图像数据进行分割.即从实际FMI资料中分离出反映孔洞、裂缝的子图像,然后用相应的方法对分割后的子图像进行分析处理,提取相应参数.本文通过研究大量图像分割算法,认为奇异点多阈值分割算法、基于过度区的分割算法、Hopfield网络方法、基于图像间模糊散度的阈值化算法是FMI图像分割的有效方法,实现了从FMI图像中将地层中有用目标,从背景中分离出来.同时,为了得到孔洞及裂缝的形状参数,本文还研究了轮廓跟踪标识边缘的方法以及根据标识的边缘进行填充的算法.在此基础上,也研究了根据目标边缘坐标计算目标的长度、宽度、圆度、裂缝密度、孔洞密度等参数的方法.上述所有方法在SUN工作站GeoFrame平台上开发成功,通过对LX45等八口井的处理结果与岩心分析数据对比表明,效果较好.

The object picking and calculating by image devision for the image log data

In order to draw the quantitative information from FMI data a important step is to segment images of FMI.That means that first of all the sub-images which can reflect cavities and cracks separated from the images of FMI,and then the subimages are processed to extract the parameters.In this paper,it has been thought that several algorithms including strangeness point multi-threshold algorithm,algorithm based on transition region,hopfield network algorithm and threshold algorithm based on fuzzy scattering degree among images were effective to separate the useful targets such as cavities and cracks from background(basement rock) by studying a lot of image segmentation algorithms.Meanwhile,in order to acquire the shape parameters of cavity and crack the method of tracking the outline of cavities and cracks to mark the edge and the stuffing algorithm according to the marked edge are studied too.On the based of these study a method to calculate the target's length,width,density of crack and density of cavities is put forward.All these method have already been transformed into programs on the GeoFrame platform in SUN workstation.By comparing the processed results of the eight wells of Tarim oilfield with the core analytical data it indicates a good effect.