Method and apparatus for encoding/decoding video using high-precision filter

1. A video decoding apparatus comprising:

a decoding unit configured to extract a quantized frequency-transformed block from a bitstream;

an inverse quantization unit configured to reconstruct a frequency-transformed block based on inversely quantizing the quantized frequency-transformed block;

an inverse transform unit configured to reconstruct a residual block based on inversely transforming the frequency-transformed block;

a prediction unit configured to generate a prediction block based on interpolating chroma sub-samples of a reference block referenced by a motion vector of a current block; and

an addition unit configured to reconstruct the current block based on adding the reconstructed residual block and the prediction block,

wherein the prediction unit is configured to

calculate non-divided values of a first set of sub-samples by multiplying chroma integer-pixels by integer numerators and summing the multiplied chroma integer-pixels,

derive the first set of sub-samples by dividing the non-divided values by a common denominator, and

derive a second set of sub-samples by interpolation from the non-divided values of the first set of sub-samples, not from the first set of sub-samples which have been divided by the common denominator, and then division by the common denominator.

2. The video decoding apparatus of claim 1 , wherein the prediction unit is configured to generate the chroma sub-samples of the reference block using a finite impulse response (FIR) filter.

3. The video decoding apparatus of claim 2 , wherein the FIR filter is a filter with 4 or more taps for calculating values of the sub-samples, based on four or more chroma integer-pixels.

4. The video decoding apparatus of claim 1 , wherein the prediction unit is configured to interpolate up to a position of a ⅛ sub-sample of a chroma component.

5. A video decoding method comprising:

generating a quantized frequency-transformed block from a bitstream;

reconstructing a frequency-transformed block by inversely quantizing the quantized frequency-transformed block;

reconstructing a residual block by inversely transforming the frequency-transformed block;

generating a prediction block based on interpolating chroma sub-samples of a reference block referenced by a motion vector of a current block; and

reconstructing the current block by adding the reconstructed residual block and the prediction block,

wherein the interpolating of the chroma sub-samples comprises:

calculating non-divided values of a first set of sub-samples by multiplying chroma integer-pixels by integer numerators and summing the multiplied integer-pixels;

deriving the first set of sub-samples by dividing the non-divided values by a common denominator, when deriving the first set of sub-samples; and

deriving a second set of sub-samples by interpolation from the non-divided values of the first set of sub-samples, not from the first set of sub-samples which have been divided by the common denominator, and then division by the common denominator, when deriving the second set of sub-samples.

6. The video decoding method of claim 5 , wherein the chroma sub-samples are generated by using a finite impulse response (FIR) filter.

7. The video decoding method of claim 6 , wherein the sub-samples are generated by the FIR filter with 4 or more taps for calculating values of the sub-samples, based on four or more integer-pixels.

8. The video decoding method of claim 5 , wherein the chroma sub-samples are interpolated up to a position of a ⅛ sub-sample.