Rock Thin Section Of Casting Porosity Image Measurement Technology Research

Oil and gas reservoir porosity is the main research object in oil geology and exploration and development of oil and gas field. Mastering porosity of oil reservoir layer is the basis to learn oil reservoir, calculate reserves, and analyze the producing status of oil field. Therefore, the porosity is the fundamental constants to be mastered for exploration and development of oil field. The porosity data is extremely important for calculating the saturation of containing oil of reservoir layer, estimating the reserve of oil, gas and other resources, and making the future development plan of country.Among the various methods of calculating porosity, researching on the thin slice of the rock cast body is a method to research on the true pore size distribution. It has intuitive and visual technology advantage in recognizing pores and calculating or evaluating reservoir formation. This paper use the theory and technology of digital image processing to research on the detection method of the pores in the cast body thin slice to obtain accurately the porosity of the reservoir formation and establish geological model.This paper uses several different image segmentation methods to detect pores in cast body chip. Firstly, the paper uses the color image histogram threshold segmentation methods, which choose HSV color space and evaluate the pore segmentation effects. Because the areas segmented by histogram threshold based on HSV color space are not totally the pore area there may be some other color background region in the segmented results so that it is necessary to do further processing and verification. Therefore, this paper presents a clustering method to detect pore in cast body thin slice, we respectively adopt RGB color clustering method and L*a*b*color K-means (K-means) clustering algorithm, and then summarizes the detection results of the pore by these two methods. Experimental results show that K-means method can get good clustering image segmentation effect. Finally, in association with the pore detection result produced by mean clustering method based on L*a*b*color space the image can be recognized more precisely by using mathematical morphological knowledge about binary image. Through calculating the ratio of pore area to total area of binary image we can obtain the porosity. The data show that the porosity obtained by other detection method and image method are basically same and have identical numerical value.