基于机器学习和图像处理的路面裂缝检测技术研究

来源期刊:中南大学学报(自然科学版)2021年第7期

论文作者:于建新 张伟光 钟靖涛 马涛 毛硕 石艺兰

文章页码:2402 - 2416

关键词:路面裂缝;卷积神经网络;图像处理;裂缝几何特性

Key words:pavement crack; convolution neural network; image processing; crack geometry characteristics

摘    要:基于机器学习,设计路面裂缝的快速检测算法,搭建卷积神经网络,对沥青路面图像进行收集和处理,分析多层感知机和卷积神经网络两类神经网络模型在沥青路面状态识别的效果。采用高精度卷积神经网络识别算法提高图像识别效率,借助混淆矩阵对比分析2类模型的识别准确率,对比空间域滤波、阈值二值化以及形态学滤波3类裂缝图像的处理方法,进行裂缝形态提取。研究结果表明:卷积神经网络模型准确率为99.75%,精度比多层感知机的高,能够对无裂缝、横向裂缝、纵向裂缝以及龟裂4类裂缝图像进行高精度识别。中值滤波算法能够有效提取路面裂缝的长度、宽度和面积,研究成果可用于路面裂缝快速检测。

Abstract: Based on machine learning, a fast detection algorithm of pavement cracks was designed, and a convolution neural network was built to collect and process the asphalt pavement image. The effect of two kinds of neural network models, multilayer perceptron and convolutional neural network, in asphalt pavement state recognition was analyzed. The high-precision convolution neural network recognition algorithm was used to improve the efficiency of image recognition. The recognition accuracy of the two types of models was compared and analyzed with the help of confusion matrix. Three kinds of processing methods of extracting crack image were compared, which were spatial domain filtering, threshold binarization and morphological filtering. The results show that the accuracy of the convolutional neural network model is 99.75%, which is higher than that of the multi-layer perceptron. It can recognize four kinds of crack images with high accuracy, including noncrack, transverse crack, longitudinal crack and alligator crack. Median filtering algorithm can extract the length, width and area of pavement cracks effectively, and the research results can be used for rapid detection of pavement cracks.

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号