QUANTITATIVE METHOD OF SHAPE PARAMETERS OF MINERAL PARTICLES BASED ON IMAGE PROCESSING

The shape characteristics of mineral particles are conventionally evaluated by thin section observation and manual identification of rock samples,the results are often subject to subjective influence. The paper presented a method for quantifying the shape parameters of mineral particles in photomicrographs of rock slices based on an image processing technology. The shape parameters of mineral particles mainly includes sphericity,convexity,aspect ratio,and roundness. Firstly,the method distinguishes mineral particles in photomicrographs using image segmentation technology,then extracts the pixel contour coordinates of the particles for discrete geometric analysis,and finally calculates the shape parameters of mineral particles. In particular,since roundness was difficult to be calculate directly and accurately,the paper proposed a two-step method to calculate the roundness based on Fourier series fitting particle contour,which includes identification the key points in corners and grouping them. The identification of the key points includes two process,which are smoothing particle contour based on Fourier series and searching for the key points in corners. The grouping of the key points also includes two process. Firstly,the key points were preliminarily divided into several groups by statistical analysis,then,the dichotomy was adopted for the precise grouping of these key points. We analyzed the effects of image resolution R_e,goodness of fitting Fourier series R² of particle profile and grouping coefficient a of key points on the calculation results of particle shape parameters. We found that(1)the results of sphericity,convexity and aspect ratio are basically not affected by R_e and R²; (2)the results of roundness are affected by R_e and R². It is suggested that the minimum circumscribed circle diameter of particles should be more than 200 pixels for roundness calculation,and there is an optimal R² range,within which the calculated roundness is close to the theoretical value; (3)when a is accurate to the tenth,a=0.1 is the most reasonable during the process of grouping of the key points.