2A1-C03 コンクリート表面画像からの対話型ひび割れ検出

概要

論文の詳細を見る
The crack is one of the most important indicators for the concrete structure diagnosis. However, it is not easy to detect cracks by the conventional image processing methods because there are a variety of difficulties such as noises, insufficient contrast and shadings. In this paper, we introduce a novel crack detection method for concrete surface image based on a percolation model. The feature of our proposed method is the semi-automated procedure in which the candidates of the crack area can be pointed by a human expert after the automatic selection through the edge extraction. As a result we accomplished a robust and reliable crack detection without an oversight. The effectiveness of our proposed method was verified by experiments on real concrete surface images.
一般社団法人日本機械学会の論文