Case studies

1 - Image in Campo Claro

The image captured in the bright field is used to segment hematite from other types of ore, such as magnetite and goethite, and from other phases such as quartz and resin.

2 - Polarized Image (CPOL)

The polarized image is used to discriminate porous hematites from compact hematites; and monocrystalline hematites from polycrystalline hematites (both belonging to the class of compact hematites).

3 - Segmentation of Hematites

Segmentation by a defined threshold (in this case, 191), to generate a binary image containing only hematite.

4 - Binary image containing all hematites, originating from the segmentation of the image in a bright field, by means of gray thresholds.

5 - Intersection between the binary image of the hematites and the polarized image, to continue in the automatic classification process.

1^The Classification:
All hematites: Compact x Porous

6 - Image of porous hematites, resulting from an automatic classification by pattern recognition using the polarized image, through a previously trained system.

1^The Classification:
All hematites: Compact x Porous

7 - Image of compact hematites, resulting from an automatic classification by pattern recognition using the polarized image, through a previously trained system.

2^The Classification:
Compact hematites: Monocrystalline x Polycrystalline

8 - Image of monocrystalline hematites, resulting from an automatic classification by pattern recognition using the polarized image, through a previously trained system.

2^The Classification:
Compact hematites: Monocrystalline x Polycrystalline

9 - Images of polycrystalline hematites, resulting from an automatic classification by pattern recognition using the polarized image, through a previously trained system.

3^The Classification:
Compact hematites: Granular x Lamellar x Lobular

10 - Granular, lamellar and lobular hematites, resulting from an automatic classification, based on the binary image, by thresholding various attributes of size and shape.