Region segmentation based radiographic detection of defects for gas turbine blades

Nondestructive testing (NDT) of mechanical structures is essential in a wide range of industries to ensure that the quality meets the design and operation requirements for safety and reliability. Since discovered by Röntgen in 1895, X-rays can be used to identify inner structures not only in medical imaging for human beings, but also in NDT for materials or objects. In the study of this paper, a direct digital radiography (DR) method is used to detect the defects in gas turbine blades. Aiming at the difficulty caused by the complex shape and uneven thickness, considering the limited size of flat-panel detectors of DR system, a region segmentation based method is presented in this paper. Using the proposed method, the radiographic sub-images can be obtained. By the reduction of scattering noise, the contrast of obtained radiographic image is enhanced and the defects are recognized. Then, quantitative and qualitative analyses are made for the detected defects. Finally, the locations of detected defects in reference to gas turbine blade are determined by the splicing of sub-images.