Abstract A method has been developed to automatically characterize the type of gemstones by optical properties of gemstones, digital image analysis and Artificial intelligence (AI). The method relies on optical properties of gemstones analysis as well as digital image acquisition and image processing. One hundred series of digital images were taken from 50 samples from different area of Iran and then transmitted to a dataset. Basic optical properties of gemstones (color, transparency, lustre, fractures, cleavages, inclusions, pleochroism, phenomenon and birefringence) are selected. Crystals of demantoid with green to yellowish-green color, crystal clusters and single crystals as large as 10–20 mm (most range from 2 to 10 mm in diameter) have been selected in the study area. The accuracy of gemstone identification using the method was %92, and %95 for agate and demantoid respectively. The method is applicable to other gemstones and it is possible to recognize the gemstone samples with the accuracy of more than 98%.

Keywords