Mineralogic 101 – Image Processing

Mineralogic 101 – Image Processing

In the previous posts on using Mineralogic we focused on analysis modes and mineral lists. One of the final critical things to consider before acquiring data is how to process the field image.

Mineralogic offers a dizzying array of potential functions that can be combined together to form an image processing recipe and applied to each field image (e.g. arithmetic, logical, morphological, segmentation and several more). The purpose of these image processing functions is to aid in the interpretation of necessary data from the fields whilst omitting unnecessary analysis areas.

A good primary example would be a basic threshold. This is the simplest of image processing recipes and ensures that the analysis mode you have chosen and mineral list you have developed is applied only to the particles (rather than the resin). It would after all be wasteful and inefficient to spend hours analysing epoxy resin.

basic-threshold

Basic threshold image processing recipe

This could be made more complicated if, for example, several threshold levels were used to split out the different phases (by BSE intensity). As can be seen below in order to split our 4 separate phases the mineral recipe has to be proportionally more detailed. However, if those individual thresholds are sufficiently geochemically homogeneous they can now be analysed using spot centroid or feature scan modes. The result is that the overall analysis time is vastly reduced (compared with mapping mode).

4-threshold

Four-threshold image processing recipe, ideal for spot centroid or feature scan analysis mode.

Another example is shown below which is a bright-phase search plus context. For this a single phase is isolated and analysed along with the immediate mineralogical context. This was developed to understand the immediate locking context around a bright phase of interest. It has since been presented at Sustainable Minerals 2016 and published in Minerals Engineering. It is particularly useful if there are only a few specific phases that are of interest (e.g. pyrite in the example below), and the information that is required is the grain size distribution of that phase, along with liberation (by free partial perimeter) and association data.

bps-plus-context

Bright phase plus context (taken from Brough et al., 2016)

 

The flexibility of Mineralogic image processing is a powerful tool and used correctly can greatly increase the data collection efficiency and overall quality of the final dataset. Futhermore, once an image processing recipe has been developed it is fully transferable between different jobs. Certainly, as with mineral lists and analysis modes, care must be taken when porting settings from previous projects but the same image processing recipe is usually sufficient for a wider range of project types, particularly when the work is something relatively straightforward like a modal mineralogy assessment.

Submit a Comment

Your email address will not be published. Required fields are marked *