基于加权融合纹理的路面裂缝检测

摘要

基于规则纹理的路面裂缝检测算法由于其自身存在的弱规则性、抗噪性差和稳定性差等问题,不能有效地检测出路面裂缝.为了解决此类问题,提出了一种基于加权融合纹理的路面裂缝检测算法.该方法采用单向多级中值滤波器滤除噪声,同时有效地保留裂缝边缘细节信息；然后融合局部二元模式特征、相对标准差和对比度三种纹理特征提取路面裂缝.实验结果表明,该算法抗噪性好,能快速有效地检测出路面裂缝.%Pavement crack detection approach based on regular texture cannot detect the pavement crack effectively for its weak regularity, bad anti-jamming ability and bad stability. In order to solve problems, a novel pavement crack detection approach based on weight texture is presented. At first, a unidirectional multi-level median filter is applied to eliminate noise while maintain the details of crack edge. Then three kinds of texture features that the local binary pattern(LBP) texture feature, the relative standard deviation and the contrast are combined to detect the pavement cracks. Experiments have been carried out and the results show that the proposed approach is robust to noise, and can detect the pavement crack effectively.