基于双层连通性检测的路面裂缝图像去噪算法

发布时间：2018-04-06 21:34

本文选题：道路工程　切入点：路面裂缝　出处：《公路交通科技》2015年11期

【摘要】：为了消除路面裂缝图像中的孤立噪声并保护裂缝边缘的细致特征,本文基于像素和裂缝子块两个层级的连通性检测设计了去噪算法,主要包括3个部分:(1)像素连通性检测,剔除连通性差的像素;(2)孤岛噪声消除算法,删除相对孤立的目标区域;(3)8×8裂缝子块去噪算法。然后,提出了算法流程并探讨参数的选择。最后,应用Visual Studio 2008和Open CV编程对本文算法、常用的中值滤波去噪法、长线段和黑色像素比例法进行性能测试,结果表明,本文算法可有效消除孤立噪声并保持边缘连续性,具有更高的准确率(85.06%)和召回率(85.80%),综合性能指标F1值比后两者分别高0.74%~19.19%和0.20%~12.06%。
[Abstract]:In order to eliminate the detailed characteristics of the pavement crack image the isolated noise and protect the edge of crack, and crack based on pixel blocks of two levels of connectivity detection design de-noising algorithm, mainly includes 3 parts: (1) pixel connectivity detection, excluding the poor connectivity of pixels; (2) elimination algorithm isolated noise, delete the target area relatively isolated; (3) 8 x 8 crack crack block denoising algorithm. Then, put forward the algorithm flow and discuss the selection of the parameters. Finally, the application of Visual Studio 2008 and Open CV programming of this algorithm, median filtering denoising method commonly used, long lines and black pixel ratio method the performance test results show that this algorithm can effectively eliminate the noise and maintain the continuity of the isolated edge, has higher accuracy and recall rate (85.06%), (85.80%) comprehensive performance index F1 value than the high 0.74%~19.19% and 0.20%~12.06%.

【作者单位】：重庆交通大学交通运输学院;西南交通大学交通运输与物流学院;西南交通大学综合运输四川省重点实验室;西南交通大学综合交通运输智能化国家地方联合工程实验室;College
【基金】：国家自然科学基金基金项目(51108391,61170041) 中央高校基本科研业务费专项资金项目(A0920502051208-99)
【分类号】：U418.6

【参考文献】