基于SVM决策树的沥青路面裂缝智能分类研究

Research on intelligent classification of asphalt pavement crack based on SVM decision tree

针对采集的沥青路面图像,首先基于改进的Mask匀光算法,使路面图像光照均匀度一致;其次通过小波变换的方法增强图像的高频系数、弱化背景部分,以增强对比度,提取裂缝目标信息;然后通过形态学操作去除背景纹理噪声及裂缝连接;最后采用SVM决策树模型实现沥青路面裂缝图像的智能分类。研究表明,改进的匀光算法较好地实现了图像均匀一致,小波变换与形态学操作能够精确地提取裂缝信息,采用的支持向量机方法识别精度满足标准要求的90%。

This paper firstly based on the improved Mask uniformation algorithm to make the uniformity of the road image uniformity.The wavelet transform method is used to enhance the high frequency coefficient and weaken the background of the image and enhance the contrast,extract the crack target information.Then remove the background texture noise and crack connection by morphological operation.Finally,the SVM decision tree model is used to realize the intelligent classification of crack image of asphalt pavement.The research shows that the improved homogenization algorithm achieves uniform image uniformity,wavelet transform and morphological operation can accurately extract crack information,and the support vector machine method uses the recognition accuracy to meet the standard requirement of 90%.