Pixel-Based Rust Severity Classification on Steel Surfaces Using HSI Features and Machine Learning Techniques

Ensuring the safety and longevity of steel structures, like bridges and buildings, depends on the effective monitoring of rust and its severity.Traditional methods of rust detection often rely on visual inspections, which can be subjective and timeconsuming.In this paper, we present a more efficient way to classify rust severity on steel surfaces using the HSI color space, along with pixel value analysis and computer vision techniques.By utilizing the different components of the HSI color model, we improve the accuracy of rust segmentation, which helps in better identifying the severity levels of rust.This method analyzes individual pixel values in images of steel surfaces, providing a detailed understanding of the rust's extent and intensity.Additionally, we conducted histogram analysis to calculate the mean and standard deviation of pixel values for assessing rust severity.Our machine learning framework, which includes Random Forest and OpenCV, processes high-resolution images to classify rust severity automatically through image segmentation and thresholding techniques.This method also incorporates the K-means algorithm with the Elbow Method to determine the optimal number of clusters and perform segmentation automatically.The results show that this approach offers a more precise and objective way of assessing rust damage according to the severity compared to traditional inspection methods.