High efficiency low complexity interpolation filters
View Patent ↗Embodiments for high efficiency low complexity interpolation filters for High Efficiency Video Coding (HEVC) are disclosed herein, specifically novel techniques for a video compression system. In order to estimate and compensate sub-pel displacements, the image signal on these sub-pel positions is generated by an interpolation process. In HEVC, sub-pel pixel interpolation is performed using filters. As described herein, the filter may have, for example, eight taps to determine the sub-pel pixel values for sub-pel pixel positions, such as half-pel and quarter-pel positions. The taps weight the integer pixels with coefficient values to generate the sub-pel signals. Different coefficients may produce different compression performance in signal distortion and noise.
1. A method for encoding video content, the method comprising:
receiving a plurality of full-pel pixel values corresponding to each line of a plurality of lines of full-pel pixels;
for each line of the plurality of lines, determining an interpolated sub-pel pixel value for each of three sub-pel pixels between two adjacent full-pel pixels, the three sub-pel pixels being a half-pel pixel, a first quarter-pel pixel and a second quarter-pel pixel between the two adjacent full-pel pixels, by:
determining a set of coefficients for an interpolation filter;
assigning a first coefficient to weight a first full full-pel pixel value in the plurality of full-pel pixel values to a first operation in the interpolation filter;
assigning a second coefficient to weight a second full-pel pixel value in the plurality of full-pel pixel values in a second operation in the interpolation filter;
assigning a third coefficient to weight a third full-pel value in the plurality of full-pel pixel values in a third operation in the interpolation filter;
assigning a fourth coefficient to weight a fourth full-pel pixel value in the plurality of full-pel pixel values in a fourth operation in the interpolation filter; and
determining the interpolated sub-pel pixel value from the first operation, the second operation, the third operation, and the fourth operation, by:
adding a first integer to a sum of outputs of the first operation, the second operation, the third operation and the fourth operation to obtain a first value; and
right-shifting the first value by a second integer to obtain the interpolated sub-pixel value; and
outputting each interpolated sub-pel pixel value for use in a motion compensation operation, wherein the set of coefficients used to determine an interpolated sub-pel pixel value for the half-pel pixel is different from the set of coefficients used to determine an interpolated sub-pel pixel value for the first quarter-pel pixel and from the set of coefficients used to determine an interpolated sub-pel pixel value for the second quarter-pel pixel.
2. The method of claim 1 wherein the set of coefficients used to determine an interpolated sub-pel pixel value for the first quarter-pel pixel is different from the set of coefficients used to determine an interpolated sub-pel pixel value for the second quarter-pel pixel.
3. The method of claim 1 wherein the set of coefficients used to determine an interpolated sub-pixel value for the half-pel pixel comprises the first coefficient having a value of −7, the second coefficient having a value of 39, the third coefficient having a value of 39, and the fourth coefficient having a value of −7.
4. The method of claim 3 wherein the set of coefficients used to determine an interpolated sub-pixel value for the first quarter-pel pixel comprises the first coefficient having a value of −5, the second coefficient having a value of 59, the third coefficient having a value of 13, and the fourth coefficient having a value of −3; and wherein the set of coefficients used to determine an interpolated sub-pixel value for the second quarter-pel pixel comprises the first coefficient having a value of −3, the second coefficient having a value of 13, the third coefficient having a value of 59, and the fourth coefficient having a value of −5.
5. The method of claim 1 wherein the first integer is 32 and the second integer is 6.
6. The method of claim 1 wherein the first integer is 128 and the second integer is 8.
7. The method of claim 1 wherein each set of coefficients for the interpolation filter comprises at least ten coefficients, a coefficient of the set of coefficients used to determine an interpolated sub-pel pixel value for the first quarter-pel pixel is equal to zero, and a coefficient of the set of coefficients used to determine an interpolated sub-pel pixel value for the second quarter-pel pixel is equal to zero.
8. A method for decoding video content, the method comprising:
receiving a plurality of full-pel pixel values corresponding to a line of a plurality of lines of full-pel pixels;
determining a first set of coefficients for an interpolation filter to interpolate a sub-pel pixel value corresponding to a half-pel position between two adjacent full-pel pixels in the line;
assigning each coefficient of the first set of coefficients to weight a respective full-pel pixel value in the plurality of full-pel pixel values in a first operation in the interpolation filter to generate an interpolated sub-pel pixel value corresponding to the half-pel position;
determining a second set of coefficients for the interpolation filter to interpolate a sub-pel pixel value corresponding to a first quarter-pel position between the two adjacent full-pel pixels in the line;
assigning each coefficient of the second set of coefficients to weight a respective full-pel pixel value in the plurality of full-pel pixel values in a second operation in the interpolation filter to generate an interpolated sub-pel pixel value corresponding to the first quarter-pel position;
determining a third set of coefficients for the interpolation filter to interpolate a sub-pel pixel value corresponding to a second quarter-pel position between the two adjacent full-pel pixels in the line;
assigning each coefficient of the third set of coefficients to weight a respective full-pel pixel value in the plurality of full-pel pixel values in a third operation in the interpolation filter to generate an interpolated sub-pel pixel value corresponding to the second quarter-pel position;
determining the interpolated sub-pel pixel value from the first operation, the second operation, the third operation, and the fourth operation, by:
adding a first integer to a sum of outputs of the first operation, the second operation, the third operation and the fourth operation to obtain a first value; and
right-shifting the first value by a second integer to obtain the interpolated sub-pixel value; and
outputting the interpolated sub-pel pixel values corresponding to the half-pel position, the first quarter-pel position and the second quarter-pel position for use in a motion compensation operation.
9. An apparatus for decoding video content, the apparatus comprising:
one or more computer processors; and
a computer-readable storage medium comprising instructions for controlling the one or more computer processors to:
receive a plurality of full-pel pixel values corresponding to a plurality of full-pel pixels of a line of in an image;
determine an interpolated sub-pel pixel value for each sub-pel pixel of at least three sub-pel pixels between two adjacent full-pel pixels by:
determining a set of coefficients for an interpolation;
assigning a first coefficient to weight a first full-pel pixel value in the plurality of full-pel pixel values in a first operation in the interpolation filter;
assigning a second coefficient to weight a second full-pel pixel value in the plurality of full-pel pixel values in a second operation in the interpolation filter;
assigning a third coefficient to weight a third full-pel pixel value in the plurality of full-pel pixel values in a third operation in the interpolation filter;
assigning a fourth coefficient to weight a fourth full-pel pixel value in the plurality of full-pel pixel values in a fourth operation in the interpolation filter; and
determining the interpolated sub-pel pixel value from the first operation, the second operation, the third operation, and the fourth operation, by:
adding a first integer to a sum of outputs of the first operation, the second operation, the third operation and the fourth operation to obtain a first value; and
right-shifting the first value by a second integer to obtain the interpolated sub-pixel value; and
output each interpolated sub-pel pixel value for use in a motion compensation operation.
10. The apparatus of claim 9 wherein the at least three sub-pel pixels comprises a quarter-pel pixel, a half-pel pixel and a three quarter-pel pixel, the set of coefficients used to determine an interpolated sub-pel pixel value for the half-pel pixel is different from the set of coefficients used to determine an interpolated sub-pel pixel value for the quarter-pel pixel and from the set of coefficients used to determine an interpolated sub-pel pixel value for the three quarter-pel pixel, and the set of coefficients used to determine the interpolated sub-pel pixel value for the quarter-pel pixel is different from the set of coefficients used to determine the interpolated sub-pel pixel value for the three quarter-pel pixel.