Road Damage Detection Based on Unsupervised Disparity Map Segmentation

TL;DR

This paper introduces an unsupervised, parameter-free algorithm for accurate and efficient pixel-level road damage detection using disparity map segmentation, achieving approximately 97.56% accuracy.

Contribution

The novel approach transforms disparity maps through energy minimization and segmentation without parameters, improving detection accuracy and efficiency over existing methods.

Findings

Pixel-level detection accuracy of 97.56%

No parameters required for damage detection

Efficient and accurate segmentation results

Abstract

This paper presents a novel road damage detection algorithm based on unsupervised disparity map segmentation. Firstly, a disparity map is transformed by minimizing an energy function with respect to stereo rig roll angle and road disparity projection model. Instead of solving this energy minimization problem using non-linear optimization techniques, we directly find its numerical solution. The transformed disparity map is then segmented using Otus's thresholding method, and the damaged road areas can be extracted. The proposed algorithm requires no parameters when detecting road damage. The experimental results illustrate that our proposed algorithm performs both accurately and efficiently. The pixel-level road damage detection accuracy is approximately 97.56%.