This paper deals with a method to estimate the 6-DOF pose of a marine vessel, which can be used in automatic mooring systems. In general, a moored vessel is affected by external forces that cause the motion of the vessel, such as wind, waves, changes in draft due to cargo loading and unloading, and changes in sea level due to tidal forces. These external forces change the pose of the vessel, and the role of the automatic mooring system is to control the motion of the vessel so that it is stable. Considering this situation, this paper proposes a vision-based 6-DOF pose estimation algorithm for marine vessel, which can produce high accurate and precise performance. The proposed method uses 2D texture information and 3D depth information together using a stereo vision system, and consists of the following steps: 2D feature extraction/description, 3D filtering, feature matching, computation of the weight of 3D correspondence pairs, and pose parameter estimation. In order to further improve the pose estimation performance in accuracy/precision point of view, this paper also proposes a two-step feature selection and a weight computation strategy through geometric matching. The proposed method is evaluated through accuracy and precision analysis for displacements for each degree of freedom.