Automated inspection of highway structures - Stage 2

The quality of data provided by visual structure inspections can vary significantly from inspector to inspector and from inspection to inspection. Improvements to the quality of the inspections are therefore desirable. Research has been undertaken with a view to developing a system for collecting a series of images covering the entire surface of the structure, and then pre-processing the images prior to delivery to the engineers. The aim is that these pre-processed images should be able to give the engineer an overall view of the condition of the structure, and draw attention to those parts of the structure which contain most defects, or defect like features. The research has investigated two main areas: image collection and display; and image analysis. The image collection and display investigation considered the use of multiple imaging positions, single imaging positions, spherical images, and mathematically transforming images to re-project them as if they were taken perpendicular to the face of interest in order to remove the effects of parallax. The image analysis research considered the possibility of using high quality images of structures to provide a consistent and quantitative assessment of condition. Sufficient information and detail was seen in the images to perform meaningful and useful image based condition assessments. The focus of the research then switched to the use of automatic image processing techniques. The research in this area has focussed on reducing the incidence of false positive reports of defects, and delivering results to the engineers which make their job simpler, quicker and more cost effective.