Video stream modifier
View Patent ↗A compressed data stream modifier 270 is disclosed. The stream modifier modifies an input data stream that may comprise audio and video data streams conforming to any one of the well-known video compression standards, for example, MPEG or AVC. The input stream is first de-multiplexed to obtain a single video elementary stream. The video elementary stream is then decoded by means of a variable-length decoder and provided to the data stream modifier. A decoding complexity of each frame is estimated by an estimator 210 and provided to a controller 250 . The controller selects a number of regions from each frame based on some psycho-visual criteria, such that a method of modification of these regions can bring the decoding complexity within an available capacity at a decoder. An efficient means of adapting a decoding complexity to an available capacity is thus obtained.
1. A non-transitory computer-readable medium encoded with data and instructions, that when executed by a computer causes the computer to:
estimate a decoding complexity required for decoding a picture from the compressed data stream; and
modify the stream when the decoding complexity of the picture approaches a capacity of the decoder, characterized in that modifying the compressed data stream, via the means for modifying, comprises modifying a method of decoding indicated in the stream;
wherein the modifying is are arranged to select a region on the basis of a heuristic psycho-visual selection criterion and modify a method of decoding for said region.
2. The computer readable medium of claim 1 , wherein the decoding complexity is based on a computational load or a memory access bandwidth for decoding the picture.
3. The computer readable medium of claim 1 , wherein the selection criterion is a proximity of the region to boundaries of the picture, a speed of movement of the region, or a measure of texture of the region.
4. The computer readable medium of claim 1 , wherein the modifying is arranged to modify a method of motion-compensated prediction of a region of a picture from bi-directional prediction to unidirectional prediction.
5. The computer readable medium of claim 1 , wherein modifying is arranged to modify a quantized transform coefficient present in the compressed data stream or arranged to select a region of a shape that adapts to an object in the picture.