Image Segmentation And Boundary Tracing Of Geological Sedimentary Facies

With the development of computer technology,computer technology is increasingly used in various aspects.In the field of geological exploration,combining computer technology with geological knowledge can make geologists more and more convenient to analyze various geological characteristics and freed geologists from complex physical labor.In the field of seismic image segmentation,geologist usually use artificial methods,but geologists have wasted a lot of effort on artificial mapping.Therefore,it is quite meaningful for geologists to introduce computer-vision related technologies into seismic image analysis.However,because of the inaccuracy of information and lack of information of the seismic facies images,the automatic image segmentation algorithms and lithologic classification methods that are commonly used today cannot obtain accurate results in seismic facies image analysis.In this thesis,the following work and innovation are made in view of information inaccuracy and information loss of seismic facies images:1.A GSH image segmentation algorithm based on human-computer interaction is proposed.The geological experience of geological experts is considered into the image segmentation algorithm by the method of human-computer interaction.Firstly,Geological experts add labels to seismic facies images through a mouse.Then the algorithm uses the super pixel segmentation algorithm to preprocess the seismic image in order to reduce the amount of data for the subsequent calculation.Then merge the nearest unlabeled super pixel by growing the labeled area input by a geological expert.Next,the unlabeled super pixels are forced into the nearest labeled area.,then we get the result of the final segmentation and this result accords with the geological experts' judgement of the seismic image.This algorithm solves the inaccurate information problem of the seismic facies image by the man-machine interaction algorithm.2.A bias field level set algorithm based on prior shape is proposed.Firstly,the GSH image segmentation algorithm based on human-computer interaction is used to divide the result map on seismic images with missing information.The geological experts then draw labels on the segmentation results,next,we use the proposed algorithm to find the edge contour of the target area.This algorithm uses the prior shape to constrain the level set to make up the missing information of the seismic image,and make the final segmentation results conform to the expectations of the geological experts.3.Based on the above two seismic image algorithms,a seismic image segmentation system was designed and implemented.And this system is introduced in detail from the perspective of system requirements analysis and module design.This thesis solves the problem of inaccurate information and lack of information in the seismic image through human-computer interaction and prior shape constraints.The results calculated by the two algorithms in this thesis not only accord with the nature of the seismic facies image itself,but also meet the geological experts' expectations.