Interphotography prediction methods and devices for video coding

Abstract

The invention relates to a method and apparatus for predicting inter-image signals for vector coding. By this method, a mixing candidate set containing MMVD candidates (mixing mechanism with moving vector differences (MVD)) is generated. When the current signal has a block size of at least one default block size, if the candidate (product in the mixing candidate set) is a bidirectional prediction determined after the mixing candidate set is generated. If the target candidate is a bidirectional prediction: the target candidate is replaced with a unidirectional prediction candidate; the unidirectional prediction candidate is contained in the mix candidate set to create a modifier of the mix candidates; and the current motion information related to the current block is encoded using the modified mix candidate set at the video encoder, or the current motion information related to the current block is decoded at the video decoder using the modified mix candidate set.