三种路面破损分类算法的仿真比较

Simulating Comparison among Three Algorithms about the Auto-Recognition of Pavement Distress

路面破损分类是制约路面破损自动检测的发展和普及的重要因素之一。本文在前文提出的破损密度方法的基础上,提炼出两种基于密度因子方法的算法——基本密度因子算法和方向密度因子算法,并通过仿真,验证了其对常见的5种路面破损类型进行分类的可行性。文中作为对照的方法是由美国Lee,ByoungJik博士提供的PROXIMITY算法。仿真结果表明1)三种路面破损分类算法中,方向密度因子算法对5种常见路面破损类型的整体识别效果最好;2)对于常规算法最难区分的块状裂缝和龟状裂缝,基本密度因子的算法的识别效果最好。这充分验证了前文中我们提出的基于破损密度因子的路面破损分类方法的可行性和优越性。

Three classification algorithms, in which density algorithm and directional density algorithm are put forward by us, and PROXIMITY algorithm by Doctor Byoung Jik Lee, are compared by simulations. Simulations show that directional density algorithm is superior in the whole than PROXIMITY algorithm, meanwhile density algorithm is inferior in the whole than PROXIMITY algorithm in recognition. But for block crack and alligator crack, which are the most difficult to be distinguished with common algorithms, our two algorithms are much better than PROXIMITY algorithm.