The invention provides a glaucoma image feature extracting method and apparatus. The method includes the steps of obtaining n characteristic values of the first and second categories according to the a priori direction vector, generating 2n-dimensional feature points according to the n characteristic values of the first and second categories and outputting the 2n-dimensional feature points. According to the glaucoma image feature extracting method disclosed in the embodiment of the present invention, a priori direction vector is obtained by calculating according to cup disc ratio vector of an eye fundus image, and a multidimensional space is then generated according to the priori direction vector. The feature points of the fundus image fall into the 2n-dimensional space, to achieve the purpose of feature extraction for an eye fundus image.