Machine Vision Approach of Bridges Crack Identification Based on the Fusion of UAV Vision and LiDAR

The application of machine vision algorithms in 3D bridges imaging and bridges inspection based on the fusion of UAV vision and LiDAR has a history of several years. Nonetheless, there is still lack of system research articles about this technology, especially for deep learning models developed to solve these problems. This article first introduces a comparison between the bridges crack detection system for classic 2D by using UAV camera images collection only and 3D by fusion between imaging information of the UAV camera technique and LiDAR technique for one crack in common that is, the original information detected by these two technologies is the scattering point information of the target. Subsequently, a convolutional neural network (CNN) is used for UAV and LiDAR imaging information feature fusion, which enhances the ability to extract damage features of bridge structures, and uses data fusion application to fusion a LiDAR and UAV detection scatter points for cracks, which improves the accuracy of bridge structure damage target detection in the real-time. The author believes that this study can provide practical guidance for the development of the next generation bridges condition evaluation system.