Multi-Modal Volumetric Imaging For Gemstone Inclusion Mapping

Abstract

The diamond industry has been rapidly developing over the last few decades. With the invention of laser cutting, the fashioning of the diamond has been made automatic and is no longer restricted by a worker's skill. However, the aspect of planning and grading rough stones is still relatively underdeveloped. Although several imaging methods have been used for diamond inspection, at present, there is no single technique able to perform comprehensive diamond clarity inspection and 3-dimensional (3D) mapping of inclusions without requiring major human intervention.

In this thesis, we developed a 3D imaging approach that is able to perform automatic inclusion mapping and clarity grading for rough diamonds. This method is designed as a turn-key solution to scanning of diamonds in the rough. It aims to be usable by general operators with no specialist training in imaging or optics, and requires minimal to no preparation of the samples prior to scanning.

The main challenge in developing an imaging method that needs no special preparation is the high refractive index of diamonds. Optical methods such as optical coherence tomography are able to identify the various internal inclusions in rough diamonds, but they are unable to accurately locate their position due to the high refractive index of diamond. Possible alternative techniques using high energy radiation probes to avoid refraction do exist, one of the example is X-ray computed tomography (XCT). It features high accuracy due to its high energy probe beam and short wavelength, but is only sensitive to inclusions which have a considerable effect on probe beam attenuation at the high energies typical for X-rays, which is often not representative of their visual impact.

In this thesis, we integrated these two 3D imaging modalities, XCT and OCT, into one multi-modal imaging system able to combine the advantages of both imaging methods. The OCT system used in this thesis is custom designed and built for this application. Instead of a single image, it takes multiple OCT sub-scans with its scan head and the sample rotating relative to each other in two rotational degrees of freedom. This method allows the system to maximise coverage of the sample volume and reduce imaging blind spots.

In post-processing, the two methods are combined, using the accurate surface envelope provided by XCT as a reference surface to stitch OCT sub-scans together. Importantly, we use the XCT surface envelope and measurements of the scan head trajectories to model and correct probe beam refraction for the corresponding OCT results. The use of software refraction correction allows this multi-modal technique to obtain refraction-free optical scans while completely avoiding sample specific preparation such as refractive index matching.

The finished system was tested to have 11 by 12 by 12 microns resolution and a maximum scan volume of 10 by 10 by 12 mm and requires no refractive index matching. We demonstrated the system's ability to achieve comparable result quality to the current industry standard method, which was confirmed by our industrial partner. The system is currently undergoing further testing and optimization in the field to prepare for commercial usage within the coming years. Show more