The advances in computerised processors and digital microscopic imaging have enabled more studies that attempt to quantify textural information from rock images. However, a texture classification system designed for geometallurgy has always been challenging, as it combines general image analysis techniques and specific texture classifications in terms of the size, shape and orientation of mineral particles. In this paper, a novel texture analysis technique using grey level radial shell association (GLRSA) and spectral angle mapper (SAM) to discriminate between different geological images is outlined. The technique can be applied in routine analysis and characterisation of mineralogical and textural attributes which may correlate with variability in mineral processing behaviours. Using this texture analysis technique, a set of small texture reference images, a texturelet reference library, has been identified from a suite of drill core images. Rock types are classified according to the proportion of each texturelet type recognised. The library is site specific.CITATION:Nguyen, K, 2013. A new texture analysis technique for geometallurgy, in Proceedings The Second AusIMM International Geometallurgy Conference (GeoMet) 2013 , pp 187-190 (The Australasian Institute of Mining and Metallurgy: Melbourne).