Microbial-induced corrosion (MIC) poses a significant challenge in various industrial settings, including space and ground flight applications. Identifying and mitigating corrosion in metallic surfaces is crucial for ensuring the structural integrity and longevity of engineering systems. Segmenting MIC allows for early detection and proactive repair, which improves safety and cost efficiency in environments such as the International Space Station. This study presents a novel baseline algorithm called CLF (clustering using local features) for MIC segmentation based on traditional Non-Deep Learning techniques. We compare our approach to two advanced Deep Learning models: DeepLabv3+ and the Segment Anything Model (SAM). The SAM model is experimented with for the first time in the context of MIC segmentation. Our results on a new MIC dataset of 154 images reveal that the SAM model excels, with an excellent Accuracy of 97.59% and a Dice Score of 71.36%. These promising outcomes highlight the potential of our methodologies and establish a strong foundation for future expansions upon the availability of further data. This research sets a critical path for further comprehensive studies in the field of MIC segmentation.