Study On Digital Core Image Segmentation Technology

Petroleum is indispensable energy of national development strategy.There are many advantages using digital core technology to describe the physical characteristics of the rock core,such as reusable,fast speed,high efficiency and more predictable physical quantities and so on,which are very important to the oil exploitation and has been get more and more attention.The accuracy of digital core image segmentation directly affect the accuracy and applicability of the digital core 3D digital modeling,and then affect the prediction and development of oil and gas fields which applying the digital core technology.Therefore,the study and application of digital core image segmentation technology has very important significance.This paper takes the digital core image as object of study,the main research contents have the following several aspects:Research the multiple threshold segmentation algorithm which is suitable for digital core image.By analyzing and comparing the principle of various transition region extraction algorithm,this paper takes the transition region extraction algorithm which is based on local complexity as the last choice.Experimental results show that this algorithm use less operation time,has high efficiency,good noise resistance and robustness.Then use histogram remaining potential function to determine the optimal clustering number,compared with the method which is based on minimum cost function,the method we selected has more efficiency.At last we use the non-maxima suppression filter to obtain the segmentation thresholds of digital core image and then segment the core image with them.The experimental results show that the result of segmentation is satisfied when using the multiple thresholds segmentation algorithm based on local complexity segment the digital core image.Research the edge extraction algorithm which is suitable for digital core image.At first we analyzed the algorithm principle of first-order differential edge detectors and second-order differential edge detectors.And do the edge detection experiment using these edge detectors.Then we focus on the feasibility and applicability of Canny edge detector and isoperimetric algorithm based on graph theory when they are applyed on the digital core image and do related experiment.The experimental results show that the edge extracted by Canny edge detector and isoperimetric algorithm based on graph theory can effectively eliminate the effect of noise in the image,get more real edge information,at the same time the positioning accuracy of edge is higher when applying the isoperimetric algorithm based on graph theory to extract the edge.Research the evaluation criterion which is suitable for digital core image segmentation algorithm.Through the analysis of the existing image segmentation evaluation criterion,this paper put forward three evaluation indexes to evaluate of digital core image segmentation algorithm,which are the quality of image segmentation results,the segmentation effiency and the automation degree of the algorithm.applying them on different core image segmentation algorithms for quantitative evaluation.The evaluation results show that the proposed multiple threshold segmentation algorithm based on local complexity for digital core image segmentation has best applicability and the highest value.