The invention belongs to the field of image data processing, particularly relates to a blood vessel registration method, system and device for an intravascular aneurysm operation, and aims at solvingthe problems that a registration image generated by an existing image registration technology is fuzzy, long in consumed time and much in interference. The method comprises the following steps: converting an acquired CTA image into an aorta blood vessel three-dimensional image, further converting the aorta blood vessel three-dimensional image into an aorta blood vessel two-dimensional image, converting a DSA image into an aorta blood vessel binary image, and obtaining a registered image through a blood vessel registration network based on deep learning, wherein the blood vessel registration network is constructed based on the regression device sub-network, the space transformation sub-network and the re-sampling sub-network. According to the invention, the problem that the DSA image is blurred in the operation can be solved, the abdominal aortic aneurysm position can be found more quickly and accurately in the current X-ray transmission image, and the requirement for assisting a doctorin the intravascular aneurysm repair operation in real time can be met.