Speedup techniques for rate distortion optimized quantization
Techniques for selecting a coding mode for an image coding process are described. Coding modes can be selected through a coding mode transition state machine, a re-quantization process, selection of an optimal transform size, by skipping some quantization parameters, or by performing motion search.
1. A method of selecting a coding mode for a coding block, comprising:
performing a first quantization process using a first quantization technique for each of M coding modes, wherein performing the first quantization process includes quantizing a prediction residual using a given quantization parameter;
based on the first quantization process, selecting M 1 coding modes from the M coding modes according to a predefined coding criterion, wherein M>M 1 ;
performing a second quantization process using a second quantization technique for the M 1 coding modes, wherein performing the second quantization process includes re-quantizing the prediction residual using the given quantization parameter, and wherein the first quantization technique has a different quantization type from that of the second quantization technique and is faster than the second quantization technique for the given quantization parameter; and
selecting the coding mode for the coding block from the M 1 coding modes based on the second quantization process.
2. The method of claim 1 , wherein the first quantization process and the second quantization process are repeated for a plurality of quantization parameters.
3. The method of claim 2 , wherein information about positions and magnitudes of quantized transform coefficients of a first quantization parameter is used during the quantization processes of a second quantization parameter.
4. The method of claim 1 , further comprising:
determining a distortion cost of a first residual quantized with the first quantization process;
determining a distortion cost of a second residual quantized with the second quantization process; and
selecting the one of the first residual and the second residual having a lower distortion cost.
5. The method of claim 1 , further comprising re-encoding the coding block after selecting the coding mode for the coding block.
6. The method of claim 5 , wherein re-encoding the coding block comprises re-encoding using the second quantization process.
7. The method of claim 6 , wherein re-encoding the coding block comprises evaluating each of the M 1 coding modes.
8. The method of claim 6 , wherein re-encoding the coding block comprises evaluating each of the M coding modes.
9. The method of claim 1 , wherein the coding mode for the coding block is selected based on the first quantization process and the second quantization process.
10. A coding apparatus, comprising:
a memory that stores a set of instructions; and
one or more processors configured to execute the set of instructions to cause the coding apparatus to:
perform a first quantization process using a first quantization technique for each of M coding modes, wherein performing the first quantization process includes quantizing a prediction residual using a given quantization parameter;
based on the first quantization process, select M 1 coding modes from the M coding modes according to a predefined coding criterion, wherein M>M 1 ;
perform a second quantization process using a second quantization technique for the M 1 coding modes, wherein performing the second quantization process includes re-quantizing the prediction residual using the given quantization parameter, and wherein the first quantization technique has a different quantization type from that of the second quantization technique and is faster than the second quantization technique for the given quantization parameter; and
select a coding mode for a coding block from the M 1 coding modes based on the second quantization process.
11. The coding apparatus of claim 10 , wherein the first quantization process and the second quantization process are repeated for a plurality of quantization parameters.
12. The coding apparatus of claim 11 , wherein information about positions and magnitudes of quantized transform coefficients of a first quantization parameter is used during the quantization processes of a second quantization parameter.
13. The coding apparatus of claim 10 , wherein the one or more processors are configured to execute the instructions to further cause the coding apparatus to:
determine a distortion cost of a first residual quantized with the first quantization process;
determine a distortion cost of a second residual quantized with the second quantization process; and
select the one of the first residual and the second residual having a lower distortion cost.
14. The coding apparatus of claim 10 , wherein the one or more processors are configured to execute the instructions to further cause the coding apparatus to:
re-encode the coding block after selecting the coding mode for the coding block.
15. The coding apparatus of claim 14 , wherein re-encoding the coding block comprises re-encoding using the second quantization process.
16. The coding apparatus of claim 15 , wherein re-encoding the coding block comprises evaluating each of the M 1 coding modes.
17. The coding apparatus of claim 15 , wherein re-encoding the coding block comprises evaluating each of the M coding modes.
18. The coding apparatus of claim 10 , wherein the coding mode for the coding block is selected based on based on the first quantization process and the second quantization process.
19. A non-transitory computer-readable storage medium having stored therein instructions that are executable by one or more processors of a coding system to perform a method for selecting a coding mode for a coding block, the method comprising:
performing a first quantization process using a first quantization technique for each of M coding modes, wherein performing the first quantization process includes quantizing a prediction residual using a given quantization parameter;
based on the first quantization process, selecting M 1 coding modes from the M coding modes according to a predefined coding criterion, wherein M>M 1 ;
performing a second quantization process using a second quantization technique for the M 1 coding modes, wherein performing the second quantization process includes re-quantizing the prediction residual using the given quantization parameter, and wherein the first quantization technique has a different quantization type from that of the second quantization technique and is faster than the second quantization technique for the given quantization parameter; and
selecting the coding mode for the coding block from the M 1 coding modes based on the second quantization process.