Ore Adhesion Particle Detection And Segmentation Algorithm Research

With study and practice in computer image processing technology become more and more mature, the computer vision are widely used in the mineral processing. Material particle size is an important feature of mineral particles, the accurate measurement of particles size is a significance indicator to subsequent processing. In order to analyze mineral particle size characterization accurately, the segmentation and separation of particle image must be done first.This paper first summarizes status of ore particle detection and analysis research in domestic and foreign, put forward research task and the main content. Introduces the basic concept of image digitization and the two process of image pre-processing:image filtering and image segmentation, using the Gaussian filter optimization algorithm for image filtering, the watershed algorithm and the OTSU optimization algorithm for image segmentation. Then describes the optimization process of the binary particle image, focused on the morphology of the particle image filtering, hole filling and eliminate the condensed state. Studied on representation of the particle parameters, gave formulas of particle area, perimeter and volume.Then discussed and given the optimization algorithm of distance transform, optimized the seed points by limit corrosion of distance transform algorithms, cut up adhesion particle partition by the greatest extent possible on the basis of avoid over-segmentation problem. Highlight and describe the adhesion segmentation algorithm:the first step do initial binary optimization of the image, the second step do the hole filling and corrosion, third step do distance transform map calculation process, fourth step do the adhesion target segmentation, fifth step do region boundary tracking and optimize again. Ultimately verified by experiment, the effect of this algorithm is obvious, it can guarantee the quality of segmentation and separation of the actual ore particle adhesion.Finally based on above theoretical study, designed a mineral particle size detection system with combination of VC++and MATLAB. The system run fast, accurate and with visually operation, achieve automatic image segmentation, boundary tracking, feature extraction functions of mineral particles, and present mineral particle size analysis in visual form at last.