The core message of this paper is to introduce a diffusion matching model that treats correspondence estimation as a denoising diffusion process within the doubly stochastic matrix space, which gradually refines a doubly stochastic matching matrix to the ground-truth one for high-quality correspondence estimation.
A semi-automatic crack segmentation tool is proposed that combines image processing techniques and user input to accurately identify and measure cracks in bridge inspection images, overcoming the limitations of fully automatic methods.