摘要： 针对现有算法检测精度不高和边缘定位不准确的问题，提出一种基于流形距离的迭代聚类路面裂缝提取算法。通过计算2个数据点之间的流形距离，设计聚类目标准则函数，利用迭代最优方法解决准则函数的优化问题，将所有数据点划分为背景和目标2个聚类，并结合图像分割算法提取路面裂缝信息。实验结果表明，该算法能稳定有效地提取出图像中的连续裂缝边缘，可用于路面裂缝的自动检测。

关键词: 裂缝提取, 流形距离, 准则函数, 图像分割, 迭代聚类

Abstract: Aiming at the problem of existing algorithm testing precision and edge position is not accurate, this paper proposes a method for pavement cracks extraction using iterative clustering algorithm based on manifold distance. The method calculate the manifold distance between two data points, and a criterion function used to express the clustering target is designed, then solve the criterion function optimization problem by iterative optimization. All the data points are divided into two clustering that the background and objectives.using image processing algorithms extract pavement cracks information. Experimental results show that the continuous cracks can be stable and effectively extracted. It is beneficial for the automatic detection of pavement cracks.

Key words: crack extraction, manifold distance, criterion function, image segmentation, iterative clustering

中图分类号: 

• TP391