2D digital human relighting method and system based on single-view video

The invention discloses a 2D digital human relighting method and system based on a single-view video, and belongs to the technical field of artificial intelligence. Acquiring a video template of the speech state of the digital person; extracting voice features corresponding to each frame of picture of the video template, and generating lip grid vertex data according to the voice features; estimating an expression coefficient and a posture coefficient of the digital human by using the lip grid vertex data and the voice features; using a pre-trained FLAME model to generate normal data of the digital human face through the expression coefficient and the attitude coefficient; the generated normal data is converted into a picture of a lip related area, and a video template original image is directly used in a lip unrelated area, and a new digital human image is obtained through fusion; and extracting an illumination component in the new digital human image, and re-rendering the current video frame based on the ill