Fast neural radiation field new view angle synthesis method and system based on self-supervised depth

The invention provides a rapid neural radiation field new view angle synthesis method and system based on self-supervision depth, relates to the field of computer vision, and mainly aims at a new view angle synthesis task based on a neural radiation field, and provides a compact NeRF model by enabling Hash Encoding to act on a Proposa network and a nerf network at the same time, so that the problem of low reconstruction speed is solved. In order to ensure that the reconstruction precision is not reduced, the reconstruction precision is improved by providing a self-supervised depth regular term and a reconstruction error loss function based on cosine similarity, and the contradiction between the reconstruction speed and the reconstruction precision of an existing method is solved. According to the method, high-speed and high-quality neural radiation field reconstruction can be supported, and guarantee is provided for tasks such as three-dimensional reconstruction and virtual reality. 本发明提出一种基于自监督深度的快速神经辐射场新视角合