For the finger-vein is under the skin, there are many inherent disadvantages for its imaging, such as biological tissues in the finger, anatomical structure, and the imaging character of skin. A novel method was proposed to solve the problem of vascular network coloboma in finger-vein IR images. Firstly, the finger-vein images were enhanced by multi-scale Gabor filter to reduce the overall image blurring. Then, the vascular skeleton network was extracted based on binarized images so as to locate the coloboma position accurately. Thirdly, the end point and the bifurcation point were extracted from the vascular skeleton network as the original point of restoration. The coloboma of the vascular skeleton network was reconstructed according to minimal path principle. Finally, the diameter of vascular network was recovered by using the Gabor directional image as a constraint. The experimental results show that this method can be used to restore local lost of vascular network and a more complete and more stable vascular network. The recognition accuracy of finger-vein images can be further improved by using the reconstructed image.