Crack Pixels Detection from Inner Side of Drain by Using Variable-sized Directional Filter

Abstract

This paper proposes a method to detect cracks in the inner side of a drain by using the edge detection techniques. Cracks can be recognized as lines with a thin width. Therefore a method for extracting edges is assumed to be useful to detect cracks. To detect a crack with an arbitrary global form, the variable-sized directional filter is applied. This filter can be set at every pixel with an optimal size and with an optimal direction. After the filter process, the threshold technique that determines a threshold to detect a crack by using its global feature is processed. The experimental results show that the proposed variable-sized directional filter and the threshold technique were effective to detect cracks in the images of inner side of drain.