Simple Procedure for Identifying Pavement Distresses from Video Images

Abstract

Pavement distress information is important to highway engineers in managing pavement networks. At the present time, there are several research groups working to develop a more efficient and unbiased method of obtaining pavement distress data. The general approach desired is to capture pavement images using video cameras mounted on a moving vehicle and then use a computer to recognize and quantify the pavement distresses from the video images. An automated pavement distress survey system developed at the University of New Mexico is described in the paper. This prototype system uses an 8‐mm camcorder, an inexpensive image‐digitizing board, and a 486 personal microcomputer. The algorithm is capable of automatically identifying longitudinal, transverse, diagonal, alligator, and map cracking. The distresses are then reported appropriately as lengths or areas according to the type of distress. The program has an accuracy in prediction of over 85% in asphalt concrete pavements and over 90% in portland cement concrete pavements. The described automated survey system is capable of accurately analyzing images captured at a vehicle speed of 24 km/h (15 mph) and below.