Quantization control for variable bit depth
The quantization parameter QP is well-known in digital video compression as an indication of picture quality. Digital symbols representing a moving image are quantized with a quantizing step that is a function QSN of the quantization parameter QP, which function QSN has been normalized to the most significant bit of the bit depth of the digital symbols. As a result, the effect of a given QP is essentially independent of bit depth a particular QP value has a standard effect on image quality, regardless of bit depth. The invention is useful, for example, in encoding and decoding at different bit depths, to generate compatible, bitstreams having different bit depths, and to allow different bit depths for different components of a video signal by compressing each with the same fidelity (i.e., the same QP).
1. A method of processing an encoded bitstream of digital symbols representing a moving image for invariance of a quantization parameter for bit depth, the encoded bitstream including the quantization parameter and quantized digital symbols that have been entropy encoded, wherein a second bit depth (M) is provided to a decoder, the method comprising: accessing, by the decoder, the quantization parameter (QP) value; mapping the quantization parameter (QP) value to a first quantization step size (QS N ) for a bit depth of N; and computing a second quantization step size (QS M ) for the second bit depth M by normalizing the first quantization size (QS N ) with respect to the most significant bit of the second bit depth M, M being greater than N, wherein the second quantization step size (QS M ) equals QSN2 (M-N) .
2. The method of claim 1 , wherein the first bit depth N equal 8.
3. The method of claim 1 , further comprising quantizing the digital symbols using a quantizer based on the second quantization size (QS M ) to generate the quantized digital symbols.
4. The method of claim 1 , further comprising de-quantizing the digital symbols using a de-quantization step based on the second quantization size (QS M ).
5. The method of claim 1 , wherein mapping the quantization parameter (QP) value to a first quantization size (QS N ) comprises applying an identity mapping.
6. The method of claim 1 , wherein mapping the quantization parameter (QP) value to a first quantization size (QS N ) comprises applying an exponential mapping.
7. A method of processing for invariance of a quantization parameter for bit depth of digital video images encoded as a bitstream of digital symbols by entropy encoding, the bitstream including the quantization parameter and quantized digital symbols, wherein a second bit depth of ten is identified to a decoder, the method comprising: accessing, by the decoder, a quantization parameter value; mapping the quantization parameter value to a first quantization step size for a first bit depth of eight; and determining a second quantization step size for the second bit depth of ten by normalizing the first quantization size with respect to the most significant bit of the second bit depth of ten, wherein the second quantization step size equals four times the first quantization step size.
8. A method for decoding a bitstream wherein the bitstream was generated by processing digital symbols representing an image, each symbol S having a bit depth N, to provide intermediate variables X, each having a bit depth N+K, where K is a function of the processing; quantizing each intermediate variable X with a quantizing step size QS N to produce a quantized data word Q, wherein QS N is a function of a quantization parameter QP, which function has been normalized to the most significant bit of the N-bit bit depth; and processing, including entropy coding, the quantized data words Q to provide an encoded bitstream, comprising:
processing, including entropy decoding, the encoded bitstream, to provide quantized data words Q,
dequantizing each quantized data word Q with a dequantization step size QS M to produce a dequantized intermediate (M+K)-bit variable X′ that approximates the intermediate (N+K)-bit variable X, wherein QS M is the same function of the quantization parameter QP as is QS N but has been normalized to the most significant bit of an M-bit bit depth, and
processing the intermediate variables X′ to produce digital symbols, each symbol S′ having a bit depth M, representing an approximation of said moving image,
wherein bit depth M equals bit depth N.