RADIOGRAPHYTo improve the speed and accuracy of defect segmentation for automated radiographic NDT, this study proposes a segmentation method based on flooding (SMBF), in which an original line-flooding algorithm and a novel adaptive thresholding method are suggested. The adaptive thresholding method is based on the growth of the flooded area. SMBF firstly detects a defect by analysing the grey level intensity profiles of a radiographic image and labels the defect found with a seed point. Then, the flooding is carried out using the line-flooding algorithm, in which water starts from a seed point and flows to the neighbour region in column-by-column order. After flooding, the threshold value is determined by detecting an abnormal growth of flooded area. In addition, comparisons of SMBF with a segmentation method based on watershed and a segmentation method using background substation are provided. The experiments demonstrate that SMBF is able to significantly reduce the segmentation time and obtain a more accurate result.