Blood vessels detection is a common task performed in numerous medical procedures. During regular medical treatments venipuncture procedures are performed for invasive medication and blood sampling. Near infrared imaging technology can be used to visualize the subcutaneous veins in cases of difficult venous access. In this paper the methods for veins centerline detection and back projection is presented. In order to highlight the suitable veins for venipuncture, the centerline of larger veins are detected and back projected to the original image. The method is applied on the near infrared images of subjects selected from four different classes of skin tone. This can be helpful to medical staff to select suitable vein for venipuncture procedures. The method is based on Frangi vesselness filter for 2D images. The artifacts like hair and small vessel-like structures are removed with morphological operations. The presented method shows consistent performance in terms of vascular structure detection for all skin tone classes. © 2015 IEEE.