Prediction of road safety risks through crack detection and structural deterioration assessment

Road surface cracks are a major contributor to vehicular accidents, particularly in high-speed and high-traffic environments. Conventional crack detection techniques that rely on grayscale imaging often fail to maintain accuracy under varying lighting conditions and in the presence of noise. To address these challenges, a robust detection methodology is proposed, based on a Gradient-based Crack Enhancement, Color Consistency, and Smoothness Regularization Model (GCSM). This model leverages Gaussian smoothing to reduce noise, gradient-based enhancement to accentuate crack features, and color consistency to effectively differentiate cracks from surrounding textures. Smoothness regularization ensures the continuity of crack patterns and minimizes false positives, enhancing the accuracy of detection. The resulting crack maps form the foundation for advanced risk analysis, directly linking crack detection to safety evaluation. The integration of crack detection with accident prediction is achieved by a hybrid model that estimates the likelihood of accidents induced by road surface deterioration. This hybrid model combines logistic regression to assess variables such as crack density, width, traffic volume, vehicle speed, and pavement condition, with a fuzzy inference system (FIS) to handle the imprecision inherent in road condition assessments. The final accident risk score is computed as a weighted combination of these components, offering enhanced prediction accuracy. Experimental results on datasets from Peshawar, Khyber Pakhtunkhwa, demonstrate that GCSM outperforms existing methods in terms of Intersection over Union (IoU), Precision, Recall, and Structural Similarity Index Measure (SSIM), with statistical significance (p < 0.01) confirmed via ANOVA. The hybrid prediction model achieves an accuracy of 88.23% and a mean squared error (MSE) of 0.042, highlighting its efficiency and robustness. This framework facilitates automated crack visualization and accident risk classification, providing valuable insights for engineers and urban planners. Future work will focus on real-time deployment and system adaptability to various road conditions, supporting intelligent transportation systems and proactive road safety management.

Text 127523.pdf - Published Version Available under License Creative Commons Attribution. Download (2MB) Official URL or Download Paper: https://www.acadlore.com/article/MITS/2025_4_4/mit...

Actions (login required)

View Item