Systems and methods for selecting a macroblock mode in a video encoder
View Patent ↗The invention is related to methods and apparatus that provide improved video encoding. Conventional video encoding techniques inadequately determine the encoding mode for a macroblock. One embodiment according to the invention advantageously includes the coding cost for motion vectors into the mode decision, thereby enhancing the quality of the selection of the mode. One embodiment further uses an L1 norm in the decision, thereby advantageously reducing the computational complexity of taking the coding cost for motion vectors into the mode decision.
1. A method of selecting a mode for the encoding of a macroblock in a video encoder, the method comprising: performing a motion search to select a motion vector; determining a residual error for the motion vector with a plurality of macroblock modes; estimating a coding cost from motion vectors for at least some of the plurality of macroblock modes; and selecting the mode for the encoding of the macroblock based on both the residual error and the coding cost associated with the mode, the step of determining a residual, error includes the steps of:
performing a sum of absolute differences (SAD) calculation;
adjusting the SAD calculation based on the coding cost for the mode;
comparing the adjusted SAD calculation for the mode to another adjusted SAD calculation for another mode; and
selecting the mode with the lower adjusted SAD calculation.
2. The method as defined in claim 1 , wherein the coding cost for an Inter macroblock with zero motion vector is zero.
3. The method as defined in claim 1 , further comprising the step of using different criteria to calculate the coding cost depending on whether the picture is a B-picture or is a P-picture.
4. The method as defined in claim 1 , further comprising the step of using different criteria to calculate the coding cost depending on whether the picture is interleaved or is progressive.
5. A method of selecting a mode for the encoding of a macroblock (MB) in a video encoder, the method comprising using both an indication for residual error and a coding cost for a motion vector for a mode to determine which mode is selected for the macroblock, the step of using an indication for residual error includes the steps of:
performing a sum of absolute differences (SAD) calculation;
adjusting the SAD calculation based on the coding cost for the mode;
comparing the adjusted SAD calculation for the mode to another adjusted SAD calculation for another mode; and
selecting the mode with the lower adjusted SAD calculation.
6. The method as defined in claim 5 , wherein the motion vector (MV) is a differential motion vector (DMV).
7. The method as defined in claim 5 , wherein the step of using an indication for residual error includes at least partially computing residual error by reusing a calculation of sum of absolute values of the residual error from a motion search.
8. A video encoder that is configured to select a mode for the encoding of a macroblock (MB), the video encoder comprising means for using both an indication for residual error and a coding cost for a motion vector for a mode to determine which mode is selected for the macroblock, the means for using both an indication for residual error and a coding cost including (a) means for performing a sum of absolute differences (SAD) calculation, (b) means far adjusting the SAD calculation based on the coding cost for the mode, (c) means for comparing the adjusted SAD calculation for the mode to another adjusted SAD calculation for another mode, and (d) means selecting the mode with the lower adjusted SAD calculation.
9. The video encoder as defined in claim 8 , wherein the means for using both an indication for residual error and a coding cost further includes means for at least partially computing residual error by reusing a calculation of sum of absolute values of the residual error from a motion search.
10. A computer program stored in a tangible medium comprising a module with instructions for selecting mode for the encoding of a macroblock (MB) when executed by a computer, the computer program including instructions for using both an indication for residual error and a coding cost for a motion vector for a mode to determine which mode is selected for the macroblock, the instructions for using an indication for residual error and a coding cost for a motion vector for a mode includes the steps of:
performing a sum of absolute differences (SAD) calculation;
adjusting the SAD calculation based on the coding cost for the mode;
comparing the adjusted SAD calculation for the mode to another adjusted SAD calculation for another mode; and
selecting the mode with the lower adjusted SAD calculation.
11. The computer program as defined in claim 10 , wherein the instructions for using an indication for residual error and a coding cost for a motion vector for a mode includes the step, of at least partially computing residual error by reusing a calculation of sum of absolute values of the residual error from a motion search.
12. A circuit configured to select a mode for the encoding of a macroblock in a fast and quality efficient manner, the circuit comprising: a circuit configured to perform a motion search to select a motion vector; a circuit configured to determine a residual error for the motion vector with a plurality of macroblock modes; a circuit configured to estimate a coding cost associated with the motion vectors for at least some of the plurality of macroblock modes; and a circuit configured to select the mode for the encoding of the macroblock based on both the residual error and the coding cost associated with the mode, the circuit configured to select the mode for the encoding of the macroblock based on both the residual error and the coding cost associated with the mode including (a) a circuit for performing a sum of absolute differences (SAD) calculation, (b) a circuit for adjusting the SAD calculation based on the coding cost for the mode, (c) a circuit for comparing the adjusted SAD calculation for the mode to another adjusted SAD calculation for another mode, and (d) a circuit for selecting the mode with the lower adjusted SAD calculation.