Asphalt and Aggregate Fluorescence Tracing Based on Sensors and Ambient Parameter Optimization

Fluorescence tracing effectively identifies asphalt stripping on aggregate surfaces, showing promise for characterizing asphalt–aggregate adhesion in pavement performance detection. However, this method’s effectiveness depends on sensor parameters and ambient conditions. This study developed a fluorescence tracing image acquisition system and employed a five-factor, six-level orthogonal experiment to optimize sensor parameters. We compared multilayer perceptron (MLP) regression, Kolmogorov–Arnold networks regression, and Laplacian sharpening for image quality assessment, with MLP proving superior. The results indicate that (1) image quality is primarily influenced by camera aperture, followed by focal length, exposure time, UV light–camera distance, and object–camera distance; (2) the optimal parameters were 100,000 ms exposure time, 8 mm focal length, 44 cm object–camera distance, aperture of 6, and 30 cm UV light–camera distance; (3) a green background with combined UV and daylight illumination in a glass box yielded the highest image quality score (0.7084); and (4) images acquired under these optimized conditions displayed fluorescence tracing and asphalt regions with superior clarity. This study optimizes the fluorescence tracing method for quantifying the adhesion between asphalt and aggregate and promotes an intellectual approach to material performance detection in pavement engineering.