System and method for performing optimized quantization via quantization re-scaling
View Patent ↗A method, system, and program product for quantizing discrete cosine transform coefficients, e.g., for MPEG compression, with minimal bit rate overhead and without using a quantization matrix. This is done by scaling a uniform quantization parameter for the entire discrete cosine transform block, defining a variety of thresholds for the quantization of discrete cosine transform coefficients below which the corresponding coefficient will be quantized to zero, and setting different normative reconstructed values for coefficients that have not been quantized to zero as the decoder will still use the original, unmodified reconstructed values as long as the corresponding coefficient is not zero.
1. A method of quantizing a plurality of transform coefficients for an encoding process, comprising:
(A) receiving the plurality of transform coefficients corresponding to a high-resolution macroblock;
(B) generating a plurality of first quantized coefficients in a circuit by quantizing said plurality of transform coefficients corresponding to said high-resolution macroblock using at least one first quantization step size;
(C) defining a zero threshold for the plurality of first quantized coefficients, wherein a first subset of the plurality of first quantized coefficients are above the zero threshold, and a second subset of the plurality of first quantized coefficients are below the zero threshold; and
generating a plurality of second quantized coefficients corresponding to said high-resolution macroblock by
(a) establishing a plurality of reconstructed transform coefficients using a second quantization step size to approximately match a subset of the plurality of transform coefficients that correspond to the first subset of the plurality of first quantized coefficients,
(b) resealing the first subset of said plurality of first quantized coefficients to the first subset of the plurality of second quantized coefficients based on the plurality of reconstructed transform coefficients and the second quantization step size, and
(c) setting to zero a second subset of said plurality of second quantized coefficients that correspond to the second subset of said plurality of first quantized coefficients,
wherein (i) each of the plurality of alternative reconstructed transform coefficients is equal to the product of the second quantization step size and one of the first subset of said plurality of second quantized coefficients and (ii) said first quantization step size is different than said second quantization step size for said encoding process.
2. The method of quantizing a plurality of transform coefficients according to claim 1 , further comprising a step of composing a bitstream carrying said plurality of second quantized coefficients and an indication of said second quantization step size such that said bit stream can be decoded by a standard decoder using said second quantization step size.
3. The method according to claim 1 , wherein said at least one first quantization step size comprises a plurality of predetermined quantization step sizes that form a standard quantization matrix not available in said standard decoder.
4. The method according to claim 1 , wherein said at least one first quantization step size comprises a plurality of predetermined quantization step sizes that form a non-standard quantization matrix that is not transmitted in a bitstream.
5. The method according to claim 1 , further comprising the step of:
selecting said second quantization step size such that said rescaling improves a rate distortion cost.
6. The method according to claim 2 , wherein said bitstream is established in a manner to allow said standard decoder to use an original and unmodified inverse quantization procedure to inverse quantize said second quantized coefficients using only said second quantization step size.
7. The method according to claim 1 , wherein said second quantization step size is less than said first quantization step size.
8. The method according to claim 1 , wherein said second quantization step size is greater than said first quantization step size.
9. The method according to claim 1 , wherein said second quantization step size is approximately a greatest denominator of said first quantized coefficients.
10. The method according to claim 1 , further comprising the step of:
rescaling a plurality of third quantized coefficients of a second block using said second quantization step size.
11. An apparatus comprising:
a first circuit for quantizing a plurality of transform coefficients for an encoding process, configured to (i) receive the plurality of transform coefficients corresponding to a high-resolution macroblock; (ii) generate a plurality of first quantized coefficients in a circuit by quantizing said plurality of transform coefficients corresponding to said high-resolution macroblock using at least one first quantization step size; (iii) define a zero threshold for the plurality of first quantized coefficients, wherein a first subset of the plurality of first quantized coefficients are above the zero threshold, and a second subset of the plurality of first quantized coefficients are below the zero threshold; and (iv) generate a plurality of second quantized coefficients corresponding to said high-resolution macroblock by
(a) establishing a plurality of reconstructed transform coefficients using a second quantization step size to approximately match a subset of the plurality of transform coefficients that correspond to the first subset of the plurality of first quantized coefficients,
(b) resealing the first subset of said plurality of first quantized coefficients to the first subset of the plurality of second quantized coefficients based on the plurality of reconstructed transform coefficients and the second quantization step size, and
(c) setting to zero a second subset of said plurality of second quantized coefficients that correspond to the second subset of said plurality of first quantized coefficients,
wherein (1) each of the plurality of alternative reconstructed transform coefficients is equal to the product of the second quantization step size and one of the first subset of said plurality of second quantized coefficients and (2) said first quantization step size is different than said second quantization step size for said encoding process; and
a second circuit configured to compose a bitstream carrying said second quantized coefficients and an indication of said second quantization step size such that said bitstream can be decoded by a standard decoder using second quantization step size.
12. The apparatus according to claim 11 , wherein selection of said second quantization step size and said composition of said bitstream are done on the fly.
13. The apparatus according to claim 11 , wherein said at least one first quantization step size comprises a plurality of predetermined quantization step sizes that form a standard quantization matrix not available in said standard decoder.
14. The apparatus according to claim 11 , wherein said at least one first quantization step size comprises a plurality of predetermined quantization step sizes that form a non-standard quantization matrix that is not transmitted in said bitstream.
15. The apparatus according to claim 11 , wherein said first circuit is further configured to select said second quantization step size such that said rescaling improves a rate distortion cost.
16. The apparatus according to claim 11 , wherein said bitstream is established in a manner to allow said standard decoder to use an original and unmodified inverse quantization procedure to inverse quantize said second quantized coefficients using only said second quantization step size.
17. The apparatus according to claim 11 , wherein said second quantization step size is less than said first quantization step size.
18. The apparatus according to claim 11 , wherein said second quantization step size is greater than said first quantization step size.
19. The apparatus according to claim 11 , wherein said second quantization step size is approximately a greatest common denominator of said first quantized coefficients.
20. An apparatus comprising:
means for quantizing a plurality of transform coefficients for an encoding process, configured to (i) receive the plurality of transform coefficients corresponding to a high-resolution macroblock; (ii) generate a plurality of first quantized coefficients in a circuit by quantizing said plurality of transform coefficients corresponding to said high-resolution macroblock using at least one first quantization step size; (iii) define a zero threshold for the plurality of first quantized coefficients, wherein a first subset of the plurality of first quantized coefficients are above the zero threshold, and a second subset of the plurality of first quantized coefficients are below the zero threshold; and (iv) generate a plurality of second quantized coefficients corresponding to said high-resolution macroblock by
(a) establishing a plurality of reconstructed transform coefficients using a second quantization step size to approximately match a subset of the plurality of transform coefficients that correspond to the first subset of the plurality of first quantized coefficients,
(b) resealing the first subset of said plurality of first quantized coefficients to the first subset of the plurality of second quantized coefficients based on the plurality of reconstructed transform coefficients and the second quantization step size, and
(c) setting to zero a second subset of said plurality of second quantized coefficients that correspond to the second subset of said plurality of first quantized coefficients,
wherein (1) each of the plurality of alternative reconstructed transform coefficients is equal to the product of the second quantization step size and one of the first subset of said plurality of second quantized coefficients and (2) said first quantization step size is different than said second quantization step size for said encoding process
means for coding configured to compose a bitstream carrying said second quantized coefficients and an indication of said second quantization step size such that said bitstream can be decoded by a standard decoder using said second quantization step size.
21. The method according to claim 2 , wherein selection of said second quantization step size and said composition of said bitstream are done on the fly.