Optimal Description Of Lithology By Statistical Methods Of Gray Level For Cuttings

As the widely application of PDC (Polycrystalline Diamond Compact) bit, it becomes difficult to identify the lithology of cuttings with traditional ways. While cutting lithology is a crucial parameter for geology description, especially in logging process, hence it is necessary to seek a new method to identify the cutting lithology. The digital image processing technology has been used in many areas, such as biomedicine, remote sensing, meteorology, forestry and agriculture etc. While in petroleum industrial and cutting logging field, its function has been rarely explored. In this thesis, the statistical methods of gray level is used to extract the features of cuttings for the purpose of providing technical support for the ongoing project of cuttings describe system in our laboratory with the hope of solving some problems of the cutting logging.The thesis altogether has four major parts. The first part begins with a detailed review of relevant study of logging which focuses on the obstacles currently confronted, foreign and domestic research status quo and several texture feature extraction methods based on the statistical methods of gray level.As the second part of the thesis, chapter two is a detail description of the photograph system developed in the laboratory. The primary work given in this chapter is to optimize the work conditions of the system. To determine the illuminating condition, images taken under natural light and incandescent light have been analyzed. It is found out that the stable light source similar to the natural sky light is a better choice under our work conditions. An annular fluorescent lamp with the CCT of 6700K becomes the first choice and a photograph system has been established. The position of the light has also been optimized. It has been found that for a better feature extraction the resolution of images should be not less than 256pixels/mm. To reach that level, a Canon EOS 20D with Canon Macro Photo Lens (MP-E 65mm f2.8 1-5X) has been applied. The quality of image has been tested and satisfying result has been obtained.The author's major work is described in Chapter 3 and chapter 4. In this part , the best description method for cutting lithology based on statistical methods of gray level, the sum and difference histograms has been determined. Three methods of histograms (gray level histogram, gray level horizontal difference histogram, and sum and difference histograms) have been applied to analyze the images to extract the features of cuttings. The results suggest that sum and difference histograms is the best method in this work. With the Bayes classifier the sum and difference histograms is used to distinguish the samples with 204 samples for training and 102 samples for testing. The accurate rate of 100% for mudstone and 98.70% for sandstone has been obtained. A classic and mature method ----gray level co-occurrence matrix is also used to distinguish the samples. The equally good results have been achieved with the accurate rate of 99.35% for mudstone and 97.71% for sandstone among 102 testing samples. Considering the other aspects, such as storage space and calculation speed, the sum and difference histograms is better than gray level co-occurrence matrix on this work.In the last part of this thesis, a general discussion of this work and some suggests of possible future developments have been given in chapter 5.