摘要
路面破损分类成为限制路面破损自动检测的普及和发展的重要因素。本文在已提出的破损密度因子算法的基础上,进一步设计出了混合密度因子,得到一种基于图像子块分布特征的路面破损识别算法。通过仿真,验证了其对常见的5种路面破损类型进行分类的可行性,并选择了另外一种路面破损分类算法来进行神经网络仿真对比。神经网络的训练样本是两组,测试样本也是两组,进行了四次仿真对比。四次仿真结果都显示混合密度因子算法有很高的识别率。
Automatic pavement crack classification has become the bottle-neck for the prevalent application of advanced automatic pavement crack equipment. Based on the concept of density factor put forward by the author, one new structure of density factor was devised and one more efficient pavement crack classification algorithm named the synthetical distress density factor was obtained in this paper. Simulation showed that the synthetical distress density factor algorithm can classify the five most common kinds of cracks (longitudinal crack, transverse crack, block crack, alligator crack and no crack) very well. At the same time, four time simulations indicate that the synthetical distress density factor algorithms are better than PROXIMITY algorithm.
出处
《交通运输工程与信息学报》
2005年第2期19-26,共8页
Journal of Transportation Engineering and Information
