Correction of Uneven Illumination in Asphalt Mixture X-Ray CT Images Based on Frequency Domain Filtering
Tengfei Nian, Shiwen Xue, Xiaoming Huang, Han Zhao, Jinguo Ge · Journal of Materials in Civil Engineering · 2025
To address the issue of uneven illumination in X-ray computed tomography (CT) images of circular sections of asphalt mixtures, this study introduces a homomorphic filtering algorithm that utilizes the polar rectangular representation of the original X-ray CT image. Through frequency domain filtering, uniform illumination of the asphalt mixture X-ray CT image is achieved. In comparison with homomorphic filtering experiments, it has been observed that directly applying frequency domain transformation to the original X-ray CT image complicates the effective distinction between the actual high- and low-frequency information present in the image. However, the construction of polar rectangular images serves as an effective solution to this issue, thereby introducing novel filtering methods for X-ray CT images of circular sections of asphalt mixtures. This study analyzes the filtering effects of each control parameter in homomorphic filtering utilizing a self-defined evaluation coefficient for uneven illumination. The results indicate that during the homomorphic filtering process of X-ray CT slices of asphalt mixtures, the low-frequency suppression parameters, and the cut-off frequency serves as the primary control factor. The enhanced homomorphic filtering algorithm effectively corrects the uneven illumination characteristics present in the X-ray CT images of continuous densely packed asphalt mixtures with a nominal particle size of up to 16 mm. Additionally, it overcomes the issue of oversegmentation of small pore structures in the image as identified by the ring Otsu’s method (OTSU) algorithm.