Complex Shadow Removal in Pavement Imagery: Leveraging Diffusion Models for Advanced Solutions

Abstract

Shadows introduce uneven illumination; obscuring crack details and causing shadow boundaries to be misinterpreted as cracks within crack detection algorithms. This study proposes a shadow removal algorithm leveraging conditional diffusion models to eliminate shadows in top-down and oblique-view pavement images. We introduce the Pavement Image Shadow Triplet Dataset (PISTD), based on the ISTD dataset for the task. Hyperparameter tuning and training are explored on 256x256 and 512x512-pixel images to determine the tradeoff between crack detection accuracy and reconstruction precision. Moreover, the two models are compared with state-of-the-art models across PSNR, RMSE, and F1 accuracy metrics. The proposed model achieves a 6% and 38% higher segmentation accuracy on top-down and oblique view images, respectively. Evaluation of the algorithm is performed with real-world shadow images with qualitative results on the images demonstrating the effectiveness of the approach.