Adaptive interpolation filters for video coding
In digital video image encoding and decoding, a filter type is selected based on symmetrical properties of the images and coefficient values of an interpolation filter are calculated based on the selected filter type. Coefficient values, filter tap-length and selected filter-type are provided in the encoded video data. Coefficient values are also calculated based on a prediction signal representative of the difference between a video frame and a reference image. The prediction signal is calculated from the reference image based on a predefined base filter and motion estimation performed on the video frame. The predefined base filter has fixed coefficient values. Coefficient values are selected from interpolation of pixel values in a selected image segment in the video frame. Symmetry properties of images can be a vertical symmetry, a horizontal symmetry and a combination thereof, so that only a portion of the filter coefficients are coded.
1. A method for encoding images in a digital video sequence, comprising:
based on symmetry of statistical properties of the images in the digital video sequence, selecting an interpolation filter used in motion compensated prediction from the group consisting of:
a first filter type in which the same filter coefficients are used for the same distances of full-pixel positions to sub-pixel positions in both the horizontal and vertical directions;
a second filter type in which the same filter coefficients are used for the same distances of full-pixel positions to sub-pixel positions in the horizontal direction;
a third filter type in which the same filter coefficients are used for the same distances of full-pixel positions to sub-pixel positions in the vertical direction;
a fourth filter type in which the same filter coefficients are used for the same distances of full-pixel positions to sub-pixel positions in the horizontal direction, and in which the same filter coefficients are used for the same distances of full-pixel positions to sub-pixel positions in the vertical direction; and
a non-symmetrical filter type;
calculating coefficient values for the selected interpolation filter based on a minimization of energy of a prediction error signal; and
providing encoded data indicative of the digital video sequence for communication to a video decoder, the encoded data comprising information indicating:
the selected filter type for the interpolation filter;
the calculated coefficient values for the interpolation filter; and
a tap length indicating a number of the coefficient values of the interpolation filter.
2. The method of claim 1 , wherein the digital video sequence comprises a sequence of video frames;
and wherein the prediction error signal is representative of a difference between a selected one of the video frames and a reference image.
3. The method of claim 1 , wherein the digital video sequence comprises a sequence of video frames;
and wherein each video frame of the sequence of video frames is partitioned in a plurality of sections, at least one section of the plurality of sections having a different symmetry property from other sections of the plurality of sections.
4. The method of claim 3 , wherein the interpolation filter comprises a filter for the at least one section and one or more other filters for the other sections.
5. An apparatus for encoding images in a digital video sequence, the apparatus comprising:
at least one processor; and
at least one memory including computer program code, the at least one memory and the computer program code configured to, with the processor, cause the apparatus at least to:
based on symmetry of statistical properties the images in the digital video sequence, select an interpolation filter used in motion compensated prediction from the group consisting of:
a first filter type in which the same filter coefficients are used for the same distances of full-pixel positions to sub-pixel positions in both the horizontal and vertical directions;
a second filter type in which the same filter coefficients are used for the same distances of full-pixel positions to sub-pixel positions in the horizontal direction;
a third filter type in which the same filter coefficients are used for the same distances of full-pixel positions to sub-pixel positions in the vertical direction;
a fourth filter type in which the same filter coefficients are used for the same distances of full-pixel positions to sub-pixel positions in the horizontal direction, and in which the same filter coefficients are used for the same distances of full-pixel positions to sub-pixel positions in the vertical direction; and
a non-symmetrical filter type;
calculate coefficient values for the selected interpolation filter based on a minimization of energy of a prediction error signal; and
provide encoded data indicative of the digital video sequence for communication to a video decoder, the encoded data including information indicating:
the selected filter type for the interpolation filter;
the calculated coefficient values for the interpolation filter; and
a tap length indicating a number of the coefficient values of the interpolation filter.
6. The apparatus of claim 5 , wherein the digital video sequence comprises a sequence of video frames;
and wherein the prediction error signal is representative of a difference between a selected one of the video frames and a reference image.
7. The apparatus of claim 5 , wherein the digital video sequence comprises a sequence of video frames;
and wherein each video frame of the sequence of video frames is partitioned in a plurality of sections, at least one section of the plurality of sections having a different symmetry property from other sections of the plurality of sections.
8. The apparatus of claim 7 , wherein the interpolation filter comprises a filter for the at least one section and one or more other filters for the other sections.