We describe the extraction of Pulmonary Artery (PA) and the Pulmonary Vein (PV) from the lungs of the male subject from the Visible Human Project (VHP), using the color images of crio-anatomical slices. We examine the characteristics of the VHP database concerning the lungs, the general approaches for vessel segmentation found in the literature and the extraction methods we propose, using state-of-the-art techniques and some contributions for their enhancement. In particular, we introduced a non-linear homogenization of the color images, based on an auto-regressive model of inter-slice differences, in order to constitute a volume suitable for 3D-segmentation methods. The segmentation itself uses a regional-based approach (3D-region growing) combined with a boundary-based approach and 3D mathematical morphology, to enhance segmentation. We also present the visualization of the 3D reconstructions from anatomical color images of the VHP database.