P—tile与直方图FCM结合的路面图像分块分割

公路养护需要用到路面图像裂缝自动检测技术，路面图像的分割是路面图像处理的关键。由于噪声等干扰因素的影响，以往利用传统的模糊C-均值聚类（FCM）算法进行路面图像分割得不到满意的结果。文章采用P-tile算法和直方图模糊C-均值聚类算法对路面图像进行分块阈值分割，既克服了传统FCM运算量大，计算速度慢的缺点，又可减少分割算法的分析范围；而且，因不同的子图有不同的阈值，可避免统一阈值的缺陷，使图像分割更加准确。实验证明，该算法能较好地分割出路面图像的裂缝。

P-tile Combined with Histogram-based FCM for Pavement Image Partitioning

The automatic detection technology of pavement image cracks needs to be used for road maintenance, and pavement image segmentation is the key of processing pavement images. Because of the influence of noise and other interference factors, using the traditional fuzzy C-means clustering （FCM） algorithm for pavement images segmentation cannot get satisfied results before. Using P-tile algorithm and Histogram-based fuzzy C-means clustering algorithm for partitioning pavement images in threshold can not only overcome the disadvantage of large and slow computing in the traditional FCM, but also reduce the analysis range of segmentation algorithm, furthermore different subimages have different thresholds, so the defect of the uniform threshold is avoided and more accurate image segmentation is made. The experiments show that the algorithm can better segment the cracks in pavement images.