Rahbin: A quadcopter unmanned aerial vehicle based on a systematic image processing approach toward an automated asphalt pavement inspection

Abstract

Automatic inspection of pavement cracking is a critical issue in pavement management systems. In this study, a quadcopter-based digital imaging system is introduced for collecting pavement surface data over a distressed area for visual conditions interpretation. An aerial evaluation was carried out using a quadcopter unmanned aerial vehicle (QUAV), which is a device equipped with a set of automatic systems. Since QUAV flies autonomously and has high maneuverability, it is potentially useful in a variety of conditions particularly the positions dangerous for surveillance and reconnaissance. The main purpose of this work is to design a multi-stage system for QUAV image analysis consisting of image processing, threshold selection, and classification stages. The images are transformed into a new domain; then, an adaptive thresholding is applied to build the pattern of transformed cracks; and finally, the polar support vector machine (the PSVM) is applied for interpretation of crack distress. The PSVM is an automation procedure based on the support vector machine (SVM) classifier defined in the polar coordinate frame. A Mixture of Wavelet modulus and three-dimensional polar Radon transform (3DPRT) are used for feature generation. We show that the PSVM method can be successfully applied to classify the crack and is capable of providing new features about cracking distress, threshold selection and classification. In order to show the applicability and efficiency of the proposed system and method, a test was conducted applying a variety of pavement distresses. The experimental results demonstrate that the applied system provides reliable output. In addition, the comparison of the derived information with the on-site manual quantifications revealed the potentiality of the QUAV and multi-stage system for future practice.