The present invention discloses a method and device for extracting pulmonary blood vessels from pectoral CT (computed tomography) images. The method includes the following steps that: n layers of inputted pectoral CT images are received, a designated image layer is obtained and is subjected to gray-scale mapping, so that a preset segmentation threshold value can be obtained; designated pixel points of a lung region are selected from the designated image layer, so that initial seed points can be obtained, 3D regional growth is performed according to the preset segmentation threshold value and the initial seed points, and a lung tissue region without blood vessels is obtained; morphological operation is performed on the lung tissue region, so that a closed lung tissue region with blood vessels is obtained, and a blood vessel threshold value is calculated; pixel points larger than the blood vessel threshold value are searched in the closed lung tissue region with blood vessels, and the pixel points are adopted as initial mark points, and an initial diffusion surface is obtained; and surface diffusion is performed in the closed lung tissue region with blood vessels according to the initial diffusion surface and the blood vessel threshold value, so that pulmonary blood vessels can be obtained. With the method and device of the invention adopted, the pulmonary blood vessels can be accurately extracted from the pectoral CT images, and therefore, doctors can be assisted to diagnose pulmonary vascular disease accurately.