Road Crack Detection and Recognition Based on Image Correction of Horizontal Camera and Convolutional Neural Network

Abstract

The total mileage of China's highway has increased substantially, ranking first worldwide with the rapid development of China's economy. Road cracks, one of the most common problems on the road, will bring great security risks if not treated and repaired in time. Therefore, how to detect and identify road cracks quickly and accurately has become the most critical link in road maintenance. This paper uses a horizontal camera to collect road image and obtain the real road image after camera calibration and image correction. Using convolution neural network, computer vision and other related algorithms, this paper studies and realizes the detection and recognition of road cracks, including identifying whether cracks exist in the detection image, and the types and sizes of cracks. This method can effectively solve complex installation problems and reduce cost. The whole algorithm adopted in the paper has a higher accuracy for image correction, crack detection, and recognition.