基于数字图像处理的路面裂缝自动分类算法

Automated Classification Algorithm of Pavement Crack Based on Digital Image Processing

为了充分掌握路面裂缝信息,给路面养护管理、路面性能评价与预测、路面结构和材料设计提供参考,对路面裂缝图像自动分类与严重程度识别进行研究。首先,将裂缝轮廓进行矢量化处理,从而分离出单个裂缝区域进行特征计算与分析,提取倾角、块度、空洞等新的裂缝几何特征;然后,选择裂缝分类特征(裂缝空洞、长宽比和倾角明显程度),基于统计阈值区分线性裂缝和网状裂缝;最后,分别根据倾角和宽度对线性裂缝进行分类和严重程度识别,根据块度特征识别块状裂缝、龟裂及其严重程度。结果表明:裂缝分类和严重程度识别结果准确有效,可在无人为干预的情况下准确、自动、实时地提取路面裂缝种类和严重程度信息。

In order to fully obtain pavement crack condition and provide a reference for pavement maintenance and management, pavement performance evaluation and prediction, pavement structure and material design, the research on automatic classification and intensity recognition of pavement cracks was conducted. First, crack contours were vectorized, thus a single crack area was separated from the others, whose crack features could be calculated and analyzed. Second, new crack features such as orientation angle, lumpiness and cavity were extracted. Then, pavement classification features were selected such as cavity, length-width ratio and significant degree of orientation angle, and based on the statistical thresholds, linear cracks and netted cracks were distinguished from each other. At last, linear crack categories and intensities were classified according to orientation angle and width respectively and block and alligator cracks and corresponding intensities were recognized on the basis of lumpiness characteristics. The resultsshow that pavement crack types and intensities are identified precisely and effectively, consequently, pavement crack automatic and real-time manner type and intensity information can be collected in an accurate, without human intervention.