TEXTURE ANALYSIS TECHNIQUES APPLIED TO RUST CLASSIFICATION IN STEEL SURFACES

摘要

The restoration of steel bridges, ships, storage tanks and steel roofs is currently characterised by the necessity of complete surface restoration which is due mainly to the use of manual surface testing methods coupled with restoration technologies. This strategy of complete surface restoration leads to long project duration, high costs and hazardous practices for both the worker and the environment. With the recent development of surface tolerant paints and some investigation of alternative medium-free surface restoration technology, it is now becoming feasible to develop novel technologies that would allow patch repair of corrode steel structures. In this paper, we propose an "Automated" nondestructive testing (NDT) method based on image processing for quantifying classification of existing and prepared steel surface condition. Such methods must be able to classify large surface areas rapidly and reliably and be suitable for automation. This paper investigates the classification of steel surfaces from a time corrosion perspective. The methods were tested and evaluated on a database of 500 images. The results provide important information about the classification parameters for rust steel surfaces analysis.