To address the need for 3D digital quality inspection of spatially distributed circular holes in aerospace manufacturing, this paper systematically investigates methods for spatial positioning and 3D reconstruction of machined circular holes based on vision measurement technology. To mitigate the influence of surface quality and texture of machined circular holes, an image extraction method for circular holes based on arc-segment grouping technology is adopted, along with a proposed effective screening mechanism. A dual-cone reconstruction and intersection algorithm is introduced to preliminarily position the circular holes in space, providing initial values for the spatial circle pose. An error equation for optimizing the spatial circle pose is established, and the Levenberg-Marquardt (LM) algorithm is used to refine the pose solution, leveraging intrinsic and extrinsic camera parameters obtained through multi-view geometry. Circular hole arrays with diameters ranging 2 – 8 mm were machined on three samples in spatial distributions. The proposed method achieved precise identification and reconstruction of circular hole edges and positions, with an average deviation of reconstructed circles from CAD models within 0.112/– 0.100 mm and a maximum deviation not exceeding 0.200/– 0.200 mm.