**5. Conclusions**

Optical image analysis characterization for all ironmaking feedstock materials needs to be of the highest quality if it is to be used to better predict downstream processing performance. Mineral and textural characterization of iron ore allows for the improved prediction of downstream processes such as beneficiation and sintering. The complex petrology of iron ore sinter also needs to be accurately characterized for sinter quality optimization. Finally, structural/textural characterization of coke is needed to best understand the connection between coke structure/texture, parent coal blend composition and final coke quality.

CSIRO optical image analysis package Mineral4/Recognition4 allows for the high-quality segmentation of phases in different materials using multi-thresholding and textural identification. In particular, it is capable of segmenting phases with the same reflectivity, but different morphology.

During coke characterization, Mineral4 successfully segments the two types of IMDC, unreacted and partially reacted, and RMDC. Segmentation of unreacted IMDC uses three comprehensive textural identification methods: bulk identification of IMDC, porous IMDC identification and identification of "washed out" IMDC, and finally combines them in one map. A similar approach, based on two methods, is used for partially reacted IMDC segmentation.

For sinter characterization, textural identification in Mineral4 allows for the segmentation of primary and secondary hematite, based on association of secondary hematite with certain other melt-precipitated phases. It also allows for the segmentation of SFCA-I from SFCA, by taking into account the micro-porous structure of the former.

In iron ore characterization, textural identification enables the segmentation of different morphologies of hematite, such as microplaty hematite and martite. Used in combination with multi-thresholding it can reliably segment dark siliceous goethite with reflectivity overlapping with that of epoxy.

This article provides detailed descriptions of textural identification algorithms utilized by Mineral4 for ironmaking-related characterization. These and similar algorithms can also be applied in other image analysis tasks where morphological segmentation is required.

**Author Contributions:** Conceptualization, E.D.; methodology, E.D. and A.P.; software, A.P. and E.D.; validation, E.D.; formal analysis, E.D. and A.P.; investigation, E.D.; data curation, E.D.; writing—original draft preparation, E.D.; writing—review and editing, E.D. and A.P.; visualization, E.D.; supervision, E.D.; project administration, E.D. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was fully funded by CSIRO.

**Acknowledgments:** The authors wish to thank CSIRO Carbon Steel Futures group staff for valuable suggestions and help during this work. We would like to express our personal acknowledgment to Michael Peterson for his useful corrections, comments and critical revision of this paper, and to Sarath Hapugoda for sharing some images.

**Conflicts of Interest:** The authors declare no conflict of interest.
