FIELD: video data processing.
SUBSTANCE: invention relates to technology for improved video signal encoding, more specifically, to managing a buffer of decoded images. Proposed is a method for decoding a video stream, involving storing pre-decoded images of the video stream in an image buffer, including a first set of images of the same time sublayer. An image can be stored in the buffer of decoded images, marked by entering an identifier into the element corresponding to the image in the list of reference images, and the identifier can be one of the following: "non-reference image" or "unused reference image". The first set of images includes a reference image of the sublayer for predicting the current image of the video stream. Herewith, the fact whether the image included in the first set of images is a sublayer non-reference image (SLNR) is determined based on the indicator presented in the slice header and/or in the image header; based on the image determined to be an SLNR image, the SLNR image is removed from the image buffer; and the current image is predicted using one or more images from at least one sublayer reference image stored in the image buffer after the SLNR image has been removed from the image buffer.
EFFECT: higher efficiency of decoding.
11 cl, 8 dwg, 4 tbl
