| 摘要 |
A filter and method of filtering modifies the computation order to accommodate horizontal symmetric filtering, and modifies the source operands while modifying the SIMD computation, so as to eliminate such heavy overhead of transposing a pixel matrix. The filter and method of filtering reformats the equations involved in the prior art to the following equations, thereby acquiring the interpolation results by reducing the required clock cycles to three cycles: <?in-line-formulae description="In-line Formulae" end="lead"?>acc=a 0 *(|p 0 +p 5 |p 1 +p 6 |p 2 +p 7 |p 3 +p 8 |)<?in-line-formulae description="In-line Formulae" end="tail"?> <?in-line-formulae description="In-line Formulae" end="lead"?>acc=a 1 *(|p 1 +p 4 |p 2 +p 5 |p 3 +p 6 |p 4 +p 7 |)+acc <?in-line-formulae description="In-line Formulae" end="tail"?> <?in-line-formulae description="In-line Formulae" end="lead"?>acc=a 2 *(|P 2 +P 3 |P 3 +P 4 |P 4 +P 5 |P 5 +P 6 |)+acc <?in-line-formulae description="In-line Formulae" end="tail"?>
|
