基于深度学习的沥青路面病害图像智能识别方法研究

Intelligent Recognition System of Asphalt Pavement Diseases Based on Image Processing and Deep Learning

针对现有沥青路面病害人工检测效率低、影响交通、安全风险大等特点,研究基于深度学习的沥青路面病害图像智能识别方法。通过对不同类型沥青路面病害整体目标区域及特殊样貌特征分别进行标记,对图像进行归一化及数据集扩充等一系列处理,以达到增强目标特征的目的。将细分的11类致灾因子作为神经网络的输入变量,通过模糊推理逻辑创建致灾因子与沥青路面病害等级的映射关系。预先处理神经网络的输入数据和后置处理神经网络的输出数据,结合引入至神经网络中的模糊推理逻辑,建立沥青路面病害综合评价模型。通过AlexNet网络识别出病害类型,并判断其危险度等级。以滇藏公路标号K170路段实际沥青路面病害为例,评判其横向裂缝危险度等级为重度,与实际情况相符,表明该沥青路面病害图像智能识别方法协同模糊神经网络综合评价模型可有效识别沥青路面病害类型并评价沥青路面病害的危险度等级。

Aiming at the characteristics of low manual detection efficiency of existing asphalt pavement diseases,traffic impact,and high safety risks,the intelligent recognition method of asphalt pavement disease images based on deep learning was studied.By marking the overall target areas and special appearance features of different types of asphalt pavement diseases,a series of processing such as image normalization and data set expansion can achieve the purpose of enhancing target features.Taking the subdivided 11 types of hazard factors as input variables of the neural network,the mapping relationship between hazard factors and asphalt pavement disease levels was created through fuzzy inference logic.By pre-processing the input data of the neural network and post-processing the output data of the neural network,combined with the fuzzy inference logic introduced into the neural network,a comprehensive evaluation model of asphalt pavement diseases was established.The input disease type was identified through the AlexNet network,and then the risk level was judged through the fuzzy neural network.Taking the actual asphalt pavement disease as an example,the danger level of transverse cracks at the section of the Yunnan-Tibet Highway marked K170 was judged to be severe,which was consistent with the actual situation,indicating that the asphalt pavement disease image intelligent recognition method combined with the fuzzy neural network comprehensive evaluation model can be used to identify the asphalt pavement disease type and evaluate the risk level of the asphalt pavement disease.