Automatic detection of asphalt pavement raveling using image texture based feature extraction and stochastic gradient descent logistic regression

Raveling is one of the critical and pervasive modes of failure observed in asphalt pavement road. Automatic detection of raveling based on image samples is a challenging task due to the complex texture of asphalt pavement. This study constructs and investigates the capability of an image processing based approach for raveling recognition. Image texture based features extracted from statistical properties of color channels and the Gray-Level Co-Occurrence Matrix are employed as input variables to characterize the state of pavement. The Stochastic Gradient Descent Logistic Regression (sGD-LR) is used to classify image samples into two categories of non-raveling and raveling based on a set of extracted features. A SGD-LR based raveling detection program has been developed in Visual C# .NET to facilitate its implementation. Experimental outcome shows that the newly constructed approach can attain a good predictive performance with a classification accuracy rate of roughly 88%. Therefore, this approach can be a helpful tool to assist transportation authorities in the task of surveying asphalt pavement condition.