The invention discloses a DSA (digital subtraction angiography) cerebrovascular image auto-segmenting method based on adjacent image feature point sets. The method includes: 1, importing a plurality of pairs of continuous DSA cerebrovascular images as source image data; 2, partitioning each pair of DSA cerebrovascular images; 3, setting an image threshold for each partitioned DSA cerebrovascular image; 4, extracting feature points on the basis of a sift algorithm; 5, acquiring feature point difference images of corresponding live images from mask images and live images in each pair of DSA cerebrovascular images subjected to feature point extraction, by means of the digital subtraction angiography; 6, extracting image feature point sets of all feature point difference images, and precisely extracting the image feature points by means of adjacent image relation; 7, subjecting the extracted image feature points to region growing to obtain corresponding cerebrovascular images. The method has the advantages that pixel information of adjacent domains is integrated through adjacent images by the image segmenting technique, feature point information extraction is more accurate, and noise is effectively decreased.