The invention provides a video coding and decoding method based on optical field rendering. The method includes the steps that first, a start frame of a new view point is calculated with the TransCAIP method; second, a scene is divided into a background area and a foreground area by combining Gaussian mixture background modeling and scene geometric calculation; third, the foreground area of the current frame of a view point 1 is calculated through a Gaussian mixture model, the foreground area is transformed into the foreground of the current frame of the new view point through a scene geometric relation, the foreground area of the last frame of the new view point is extracted through the Gaussian mixture model, the foreground area of the last frame of the new view point is updated through the foreground area of the current frame of the new view point, and the current frame of the new view point is obtained; fourth, calculation is performed continuously K-1 times, statistics is performed on the distribution of levels where foreground area points are located in the circulation; fifth, when the number of frames of the circulation exceeds K, the levels of the screen are divided adaptively according to the foreground point distribution in the statistics; the first step is executed again. Through the method, details of images are made more clear and natural, the rendering effect is good, and the data amount is large.