关键词: 道路工程, 沥青混合料, 数字图像处理, HSI颜色空间, 图像亮度增强, 图像分割

Abstract: In view of the inaccurate segmentation of aggregate particles in the image preprocessing of paving mixture under low illumination conditions, the retinnex brightness enhancement method in HSI color space of digital image was utilized. Through the digital images of asphalt mixture collected under two kinds of light conditions, the traditional RGB color space and HSI color space image brightness enhancement methods were used respectively for preprocessing. Compared with the traditional RGB color space processing method and HSI color space image processing method, the segmentation accuracy of gray level co-occurrence matrix of image preprocessing and the proportion of aggregate area ratio above 9.5 mm in binary image after image preprocessing was studied, which was suitable for the image processing of paving asphalt mixture under low illumination. The research results show that: through the proposed HSI color space image brightness enhancement preprocessing algorithm, the entropy and energy of gray level co-occurrence matrix under low illumination condition are close to those under sufficient light condition, and the error between aggregate area ratio and extraction screening test results is small. The digital image preprocessing method of paving asphalt mixture in HSI color space has high capability of image restoration and accuracy of aggregate segmentation recognition under low illumination conditions, which is suitable for the application of asphalt mixture image preprocessing under dark light conditions in practical engineering.

Key words: highway engineering, asphalt mixture, digital image processing, HSI color space, image brightness enhancement, image segmentation