ABSTRACT
Rut depth is one of the main indices in road maintenance. The current practice to evaluate rut depth is by manual inspection or by an irradiator and a special camera installed on an inspection vehicle. A new method to estimate rut depth by image processing that only needs an in-vehicle camera is presented. Images extracted from video captured by a camera are transformed into top-view images, and then two consecutive images are compared through feature matching techniques to evaluate the image distortion due to rutting. Rut depth information can thus be obtained from 2D camera images. This method is applied to road surfaces and estimation results are compared with true values. The results show that rut depth can be evaluated with good accuracy in this inexpensive and efficient way.
