Slag Detection From Molten Material Using Visible Images and the SIFT Algorithm

Identifying slag from molten material is one of the key challenges in the steel production industry. Today, solving this challenge using intelligent methods based on image processing has gained attention. Given the importance of this issue, this study proposes a lightweight feature-based image processing approach based on the SIFT algorithm to detect slag from molten material. In this algorithm, the received image is first converted to grayscale, and then the Haar filter is applied to it. The LL portion of the image, which contains the least details and the most information, is selected, and the SIFT algorithm is applied to it. Finally, features are extracted using BRIEF. The performance of the proposed method is evaluated using parameters such as Sensitivity, Precision, Accuracy, and IoU, yielding values of 0.9, 0.9, 0.9, 0.82, and 0.9, respectively, indicating the favorable performance of the proposed algorithm.