Defect detection, which plays a positive role in reducing or avoiding accidents, is the key to ensuring the operational safety of trains. In this paper, a combining computed framework is designed for defect detection in digital radiography (DR) images of castings on a railway freight
car, combining the two classical geometry active contour models: CV and LBF. Firstly, the LBF model is used to extract workpiece regions including defects, so that the strong edge effect can be eliminated on detection. Secondly, after enhancing the local workpiece image contrast, the CV model
is chosen to segment the defect regions. Finally, after removing pseudo-defects by setting a threshold, the geometric parameters such as centroid and area of defect regions are obtained. The experimental results demonstrate that this combining method can accurately segment defect regions in
the DR image and extract useful information of geometric parameters.