Research On Crack Detection Algorithm For Oil Rock Core Industrial CT Image

Since oil spreads in sedimentary rock crevices, the research on cracks detection of sedimentary cores is quite significant for geological survey and oil exploration. Industrial CT is an excellent kind of non-destructive testing technology, which can achieve the test about cracks on oil sedimentary core and obtain CT images. From the processing and analysis on CT image obtained, we can get the distribution of cracks inside, and then measure their geometric parameters. The segmentation of cracks on CT image is an important part of the analysis and measurement on cracks. In the industrial CT images of oil sedimentary core, a great deal of background area is often contained, but gray level varies greatly between background region and core region.The tiny crack area is often contained inside the core region, and the difference of gray level between crack and workpiece is so small, therefore, the big gray difference between the background area and the workpiece affects the evolution process of curve segmentation algorithm, as a result, the background area and separate work area are successfully separated, however the result about crack region segmentation is not satisfactory. Therefore accurate crack segmentation on CT images of oil sedimentary core is quite difficult to get.Based on above issues, the main studies in this paper are as follows:Level set algorithm is a kind of image segmentation algorithm based on the theory of partial differential equations; it is also an effective segmentation method in today’s research. According to the need of crack region segmentation and the characteristics of oil sedimentary core industrial CT image, the existing level set segmentation algorithm is improved, and the accurate segmentation result has been achieved. Firstly, CV model is applied to get the coarse image segmentation result, and then the gray level of the background region is changed in order to make all pixels in background are given a new level- the average gray level of the pixels in work area, while median filtering denoising processing is ongoing. Finally, the RSF model is applied in crack region segmentation. The experimental results are satisfactory, indicating that the improved level set algorithm for image segmentation is effective for oil sedimentary core CT image.In addition, in this paper, a set of segmentation results about core crack CT images are worked out. Based on these results, 3D renderings of the crack segmentation are achieved. In order to deal with the images with serious noise pollution, a new de-noising method is used, and the segmentation results show that the method is effective.In the part of parametric analysis, an important parameter sigma in RSF model is studied. In experiment, the influences that sigma makes on segmentation evolution time and result, and the select of sigma value according to different size images are studied.