Authors: Assistant Professor G.Sudheer Kumar, Deepthi Reddy Kethireddy
Abstract: Wire ropes are essential load-bearing components in construction machinery, elevators, cranes, and other heavy industrial systems, where their structural integrity directly influences operational safety and reliability. Detecting defects such as wire breakage at an early stage is critical for preventing equipment failure and minimizing the risk of severe accidents. However, conventional inspection procedures are predominantly performed through manual visual examination, making the inspection process dependent on operator experience, environmental conditions, and available working time. These limitations often lead to inconsistent inspection quality and reduce the reliability of damage assessment. This paper presents an image processing-based framework for automated wire rope damage detection that minimizes dependence on human expertise while providing a practical and efficient inspection solution. The proposed methodology investigates two complementary image analysis techniques: an Autoencoder-based anomaly detection model and a Gabor filter-based feature extraction approach for identifying wire breakage and other surface abnormalities. The Autoencoder is trained exclusively using normal wire rope images to learn the visual characteristics of undamaged ropes, allowing abnormal regions to be identified through reconstruction error analysis. Additionally, a Convolutional Autoencoder (CAE) is evaluated to examine its capability for anomaly reconstruction, while connectivity analysis combined with Otsu thresholding is employed to improve the detection of wear-related defects. The Gabor filter method exploits local frequency characteristics of wire rope textures to identify minute wire breakage with high sensitivity. Experimental evaluation demonstrates that the Autoencoder-based approach is effective for identifying wear regions through connectivity analysis, whereas the Gabor filter successfully detects fine wire breakage that is difficult to identify using reconstruction-based methods alone. The proposed framework provides a practical, cost-effective, and image-based inspection strategy that supports safer and more reliable wire rope monitoring in construction environments.
