| 摘要 |
A "multi-resolution signal renderer" provides a computationally efficient process for generating reduced-resolution versions of a hierarchical transform coded digital signal from the encoded coefficients of that signal. The multi-resolution signal renderer begins by decoding the transform coded signal up to the highest hierarchical resolution not exceeding the desired signal resolution. The multi-resolution signal renderer then operates in one of two modes to generate reduced resolution signals. First, to generate signals at coded hierarchical resolution levels, DC coefficients of the transform coefficients are scaled, and then lowpass filtered to reduce aliasing in the signal. Second, to generate intermediate resolution signals between coded hierarchical resolution levels, the multi-resolution signal renderer retrieves the encoded coefficient blocks for the next highest resolution coding level, lowpass filters those blocks in the transform domain, performs an inverse transform, and then downsamples the resulting signal to the desired resolution.
|
