Software for a particle-size analyser based on image analysis techniques

摘要

This dissertation describes a technique of obtaining particle-size measurementudusing 2-dimensional image analysis. Images of coarse material onudrun-of- mine conveyor belts may be processed using the methodology describedudand size-calculation of particles is possible to a high degreeudof accuracy. The processing involves a histogram modification techniqueudfor performing image segmentation, a rule-driven procedure for mergingudregions and a region-splitting method to resolve particles that are joinedudtogether. Images of varying brightness and contrast may be processedudwithout adjustment of any threshold levels, hence lending reliability andudrobustness to the approach. Particles may also be resolved from very indistinctudbackground features formed by the finer material that is usuallyudpresent. The work includes the software processing necessary to performudparticle recognition on a given image frame providing a foundation onudwhich a real-time instrument may be developed.