基于稀疏自编码的路面裂缝检测  

Pavement Crack Detection Based on Sparse AutoEncoder

作　　者：钱彬[1] 唐振民[1] 徐威[1] 

机构地区：[1]南京理工大学计算机科学与工程学院,江苏南京210094

出　　处：《北京理工大学学报》2015年第8期800-804,809,共6页Transactions of Beijing Institute of Technology

基　　金：国家自然科学基金资助项目(61305134)

摘　　要：针对传统路面裂缝检测系统在复杂纹理背景噪声下检测效率低,易造成漏检、错检等现象提出了一种基于稀疏自编码的裂缝自动检测方法.该方法首先采用一种基于各向异性的检测算法进行裂缝子块的初步筛选,经过稀疏自编码提取出特征后由softmax分类器进行训练和分类,最后由张量投票算法进行空间加强和去噪从而得到裂缝信息.实验结果表明,文中提出的算法在无人工干预的情况下能够有效检测出图像裂缝区域,相比传统检测算法具有更高的检测精度和抗干扰能力.Traditional pavement crack detection system can hardly detect cracks accurately due tothe complicated background noises over the pavement surface. So a novel crack detection methodbased on sparse autoencoder was proposed. Firstly, an anisotropy detection algorithm wasadopted to select the potential crack patches. Then the features of crack patches were extractedthrough sparse autoencoder and then trained by softmax to classify. Finally, benefited by thetensor voting based spatial enhancement, the cracks were extracted after noises-removing.Experimental results show that the proposed method can meet the requirement of crackdetection. It is superior to other traditional methods with high accuracy and robustness.

关 键 词：路面裂缝 稀疏自编码 各向异性 张量投票 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]