In this paper, we propose a monocular 3D pose estimation method which explicitly takes into account the angles between the camera optical axis and bones (camera-bone angles) as well as temporal information. The proposed method combines a 2D-to-3D-based method, which predicts a 3D pose from a sequence of 2D poses, and convolutional neural network (CNN) and includes novel regularization loss to enable the CNN to extract camera-bone-angle information. The camera-bone-angle and temporal information suppress ambiguity of 2D-to-3D-based methods where the same 2D pose can be mapped to multiple 3D poses. Experiments on the Human3.6M and MPI-INF-3DHP datasets showed that the proposed method improved the performance by 5.1 mm and 2.1 mm in terms of mean per joint position error (MPJPE) respectively.