Joint image and acceleration-based algorithms for pavement recognition

In the road surface recognition technology of automobiles, to make-up for the deficiency that image recognition is easily affected by the light of the surrounding environment, this question proposes a road surface recognition method that utilizes the vertical acceleration of vehicles to assist image recognition. Based on image recognition with the residual convolutional neural network as the core, the collected acceleration data are statistically analyzed using descriptive statistics, after which the two methods are combined for road surface recognition. After experimental testing, the joint algorithm improves the overall pavement recognition accuracy.