| 摘要 |
A method of optimising a quantisation level for compressing an input video signal includes performing a transformation on the video signal to a spatial frequency domain to determine coefficients of the transformed signal representative of the video signal and determining a quantisation error introduced with each of a range of quantisation values, operating on the coefficients to determine a selected quantisation value corresponding to a local minimum quantisation error. The selected quantisation value is used for quantisation of the transformed signal to output a compressed video signal. The selected quantisation value may also be used for inverse quantisation of the compressed video signal. An inverse of the spatial frequency domain transform of the inverse quantised compressed signal forms an uncompressed signal and the uncompressed signal is used, together with the input video signal, for motion compensation. |
