A registration method is proposed for 3D reconstruction of an indoor environment using a multi-view camera. In general, previous methods have a high computational complexity and are not robust for 3D point cloud with low precision. Thus, a projection-based registration is presented. First, depth are refined based on temporal property by excluding 3D points with a large variation, and spatial property by filling holes referring neighboring 3D points. Second, 3D point clouds acquired at two views are projected onto the same image plane, and two-step integer mapping enables the modified KLT to find correspondences. Then, fine registration is carried out by minimizing distance errors. Finally, a final color is evaluated using colors of corresponding points and an indoor environment is reconstructed by applying the above procedure to consecutive scenes. The proposed method reduces computational complexity by searching for correspondences within an image plane. It not only enables an effective registration even for 3D point cloud with low precision, but also need only a few views. The generated model can be adopted for interaction with as well as navigation in a virtual environment.