Gravity-aided inertial navigation is a key technology for the navigation and positioning of underwater vehicles, while the matching area selection algorithm provides guidance for the matching process. This paper first proposes a gravity field data processing method by using Mercator projection and Delaunay triangulation, which transforms the traditional grid data structure into a gravity field triangular network structure, which is conducive to fully exploring the geometric and spatial characteristics of the gravity field. In addition, the spatial characteristics of the normal vectors of the gravity field triangular surface and the navigation direction are introduced for clustering, then a matching area selection algorithm based on geometric vectors is proposed, which can select the strong matching area according to the real-time sailing direction of the underwater vehicle. Simulation experiments show that the algorithm proposed in this paper can construct a radar chart of the matching area numbers in the gravity field on the sailing direction, guiding the underwater vehicle to navigate to areas with rich gravity anomaly information for matching and positioning.