MPI new viewpoint synthesis method based on global feature modeling

The invention relates to a multi-plane image (MPI) new viewpoint synthesis method (TransMPI) based on global feature modeling, in particular to a multi-plane image (MPI) new viewpoint synthesis method based on global feature modeling. According to the method, an MPI generation network firstly utilizes a 3D encoder to capture local spatial features among a plurality of depth planes, the prediction capability of an MPI depth plane occlusion area is improved, a Transform self-attention mechanism is introduced in order to overcome the limitation of a 3D convolutional neural network (CNN) on global semantic information learning, TransMPI performs global feature representation modeling by combining a Transform encoder, and the prediction capability of the MPI depth plane occlusion area is improved. A long-distance dependency relationship is established in a global space. Experimental results show that TransMPI further improves the reasoning quality of MPI scene representation by learning global features and local f