| 摘要 |
The present invention relates to a hybrid method for texture representation and, in particular, to image and video data compression based on this method and corresponding encoder and decoder apparatuses. An input signal is separated in the frequency domain into two sub-band components. The low-frequency component is encoded faithfully by means of a conventional image/video encoder. The high-frequency component, on the other hand, is analyzed so as to compute representative texture parameters. Instead of faithfully encoding the high-frequency component, only the computed texture parameters are stored or transmitted. Upon decoding, the low-frequency component is reconstructed, whereas the high-frequency component is replaced by a similar-looking texture that has been synthesized in accordance with the texture parameters. Both the reconstructed low-frequency component and the synthesized high-frequency component are then merged in order to generate the output signal. |
