Study On Stereo Microscopic Image Segmentation Technology Of Cores

The stereo microscopic images of cores are new section images collected by stereomicroscope,which can intuitively reflect the 3d morphology and distribution of cores In order to reduce the influence of noise and improve the computing efficiency,segmentation accuracy and adaptability based on the detailed study of image segmentation technology and advanced computer technology,this paper studied the segmentation algorithm of rock particles and crude oil,and realized the description on rock particles and crude oil from qualitative to quantitative or semi-quantitative.The main research contents were listed as follows.Analyzed complex distribution characteristics of rock particles in microscopic images,and designed segmentation algorithm used for rock particles.In order to reduce the influence of the high light which can not be eliminated by traditional method on subsequent operation,we used nonlinear adaptive transformation method to compensate light and applied local minimum method to eliminate high light before images segmentation process.During images segmentation process,in order to improve operational efficiency of algorithm,this paper pre-segmented images through secondary watershed algorithm and distributed watershed 0 to obtain region blocks which can be regarded as nodes of Ncut to reduce the number of nodes.In order to improve segmentation accuracy,we used average spatial position and average color representation of all Rixels in the region to represent regional block spatial position characteristics and color characteristics.Moreover,the optimal similarity weight matrix was reconstructed by introducing edge gradient amplitude feature.In order to improve the algorithm adaptability,we used the standard deviation of color feature,spatial position and edge gradient to replace sensitive difference parameter,which reduced the influence of subjective factors.In order to improve the stability of algorithm,we chosen region block as far as possible based on the similarity matrix feature as K-means initial clustering center,and the uncertainty that the random selection may cause was reduced.Experimental results showed that this algorithm can basically realize pretreatment of rock particles,and could be applied to the statistical operation of geometric characteristics of rock particles.Due to color and distribution diversity of the crude oil,this paper proposed a color similarity segmentation algorithm used for crude oil.We determined the crude oil main color by interactively defining,and as for the relevant color,we selected the RGB value of all pixels of image.Then we discussed the 0 component appears in its RGB value,and analyzed the crude oil level.An analysis system was presented for core stereo microscopic images based on Matlab GUIDE,and it can segment rock particles and crude oil,describe their parameters semi-quantitatively,and output the standard observation analysis report.