Discrete X-ray tomographic reconstruction for fast mineral liberation spectrum retrieval

• X-ray tomography is a promising tool for un-biased mineral liberation studies.
• A reconstruction method that uses few projections for generating a slice is proposed.
• An original approach is used and the reconstruction is solved by binary programming.
• The scanning-time is cut and mineral liberation data is generated in shorter time.
• Results of numerical experiments are compared to traditional reconstruction methods.

In mineral beneficiation, the mineral liberation spectrum of the plant feed conveys valuable information for adjusting operations, provided that it is available in minutes from particulate sampling. X-ray micro-tomography is the only technique available for unbiased measurement of composite particle composition (on a 3D basis). The bottleneck of current micro-tomographic systems is the X-ray scanning time (data acquisition) rather than the slice reconstruction time (data processing). An algorithm capable of reconstructing tomographic slices of composite mineral particles from a limited number of radiographic projections, thus significantly reducing the overall measurement time, is presented and demonstrated with numerical examples. The algorithm is cast around the discrete algebraical reconstruction technique and requires less than one tenth of the projection data needed by the currently used filtered back-projection methods, thus allowing a dramatic reduction of the scanning time.