Missions to small bodies typically involve an initial characterization phase, where a 3D shape model is constructed for the purposes of vision-based relative navigation and estimation of the dynamical properties of the body. Current state-of-thepractice methods for shape reconstruction, such as Stereophotoclinometry (SPC), rely heavily on human-in-the-loop processes executed on the ground, thus constraining the level of autonomy that can be achieved by these missions. This work details a visual simultaneous localization and mapping (VSLAM) based autonomous shape reconstruction algorithm for the estimation of the dynamical properties and for the construction of a preliminary shape model of a small celestial body. The proposed method is applied to both simulated and real image data and is shown to accurately reconstruct the 3D shape of the target asteroid.