Detecção automática de vazios em isoladores poliméricos por tomografia industrial 3 D Automatic detection of flaws in polymer insulators using 3 D industrial tomography

This work presents a methodology for the automatic detection of flaws in polymer insulators using three-dimensional industrial computed tomography (CT), as well as results obtained in the context of power distribution networks. The CT slices were reconstructed using 180 digital radiographs (projections) acquired by a high resolution system (pixel dimension of 50 μm × 50 μm, a-Si). For the reconstruction of 3D CT, 50 μm wide bitmap slices were used. The Marching Cubes algorithm was used to perform the 3D reconstruction, using the Visualization Toolkit (VTK) library and the Java programming language (64 Bits Linux platform). Nine features were obtained from the reconstructed three-dimensional objects for the neural networks training. Results showed to be satisfactory.