Encoding and decoding perceptually-quantized video content

1. A method to improve coding efficiency and perceptual uniformity for high dynamic range (HDR) video, the method comprising:

accessing a first video signal in a first color space; and

applying one or more color transformations to the first video signal to generate an output video signal in a perceptually-quantized opponent color space,

wherein applying one or more color transformations to the first video signal to generate the output video signal further comprises:

applying a first color transform to the first video signal to generate a second video signal in a linear primary color space;

applying a color cross-talk transformation matrix to the second video signal to generate a third video signal in the linear primary color space;

applying a perceptual quantizer to the third video signal to generate a fourth video signal in a perceptually quantized primary color space; and

applying a second color transform to the fourth video signal to generate the output video signal in the perceptually-quantized opponent color space.

2. The method of claim 1 , wherein the first color space is gamma-coded RGB (RGB gamma) or perceptually quantized RGB (RGB-PQ).

3. The method of claim 1 , further comprising:

applying a linearization function to the first video signal to generate a linearized first video signal and applying the first color transform to the linearized first video signal.

4. The method of claim 1 , wherein the perceptually quantized color space is generated according to the SMPTE specification ST 1084.

5. The method of claim 1 , wherein the linear primary color space is the LMS color space.

6. The method of claim 1 , wherein the perceptually-quantized opponent color space is the perceptually quantized IPT color space (IPT-PQ).

7. The method of claim 1 , wherein the color cross-talk transformation matrix comprises

C

=

[

1

-

2

⁢

c

c

c

c

1

-

2

⁢

c

c

c

c

1

-

2

⁢

c

]

wherein c is a constant.

8. The method of claim 7 , wherein the value of c is determined as a function of the amount of valid code-words in the chroma components of the perceptually-quantized opponent color space.

9. The method of claim 8 , wherein the value of c is determined as the value for which the first derivative of E(c) is maximized, wherein E(c) denotes a function of valid code-words in the chroma components of the perceptually-quantized opponent color space as a function of c.

10. The method of claim 9 wherein the perceptually-quantized opponent color space comprises the IPT-PQ color space where P and T comprise the chroma components.

11. The method of claim 1 , further comprising applying a reshaping function to the output video signal to generate a reshaped output function and compressing the reshaped output function with a video encoder to generate a coded bit stream.

12. The method of claim 11 , wherein the reshaping function is non-linear for a luma component of the output video signal and linear for one or more chroma components of the output video signal.

13. The method of claim 12 , wherein the reshaping function for a chroma channel (P) of the output video signal comprises an input-output function given by:

s

i

P

=

clip

⁢

⁢

3

⁢

(

⌊

w

P

⁢

C

H

I

-

C

L

I

v

H

I

-

v

L

I

⁢

(

v

i

P

-

v

L

P

+

v

H

P

2

)

+

mid

+

0.5

⌋

,

0

,

max

)

,

where mid and max denote a middle-range value and a maximum value of the input to the reshaping function, v i P denotes an input pixel value to the reshape function, s i P denotes the corresponding output value, w p is a weighting constant, v L P , v H P , v L I and, v H I are low and upper limits of the output video signal for the luma (I) and the chroma (P) channels, and C L I and C H I are lower and upper bounds for the luma channel values of the reshaped output function.

14. The method of claim 1 , further comprising compressing the output video signal with a video encoder to generate a coded bit stream.

15. In a decoder, a method to generate an uncompressed HDR video signal, the method comprising:

decoding a coded bit stream with a decoder to generate a first HDR signal in a perceptually-quantized opponent color space;

applying a first set of one or more color transformations to the first HDR signal to generate a second video signal in a first linear primary color space;

applying the inverse of a color cross-talk transformation matrix to the second video signal to generate a third video signal in the first linear primary color space; and

applying a second set of one or more color transformations to the third video signal to generate a fourth video signal in a second linear primary color space.

16. The method of claim 15 , wherein the perceptually-quantized opponent color space comprises the IPT-PQ color space.

17. The method of claim 15 , wherein the first linear primary color space comprises the linear LMS color space and the second linear primary color space comprises the linear RGB color space.

18. The method of claim 15 , wherein the inverse of the color cross-talk transformation matrix comprises the inverse of

C

=

[

1

-

2

⁢

c

c

c

c

1

-

2

⁢

c

c

c

c

1

-

2

⁢

c

]

,

wherein c is a constant.

19. The method of claim 18 , wherein the value of c is signaled from an encoder to the decoder.

20. The method of claim 15 , further comprising applying a backward reshape function to the first HDR signal to generate a reshaped first HDR signal; and applying the first set of one or more color transformations to the reshaped first HDR signal to generate the second video signal.

21. An apparatus comprising a processor and configured to perform the method recited in claim 1 .

22. A non-transitory computer-readable storage medium having stored thereon computer-executable instructions for executing a method with a processor in accordance with any claim 1 .