Method and apparatus for processing high dynamic range images
An input HDR image may be represented using an SDR image and a modulation value. For the SDR image to be viewable and representative of the HDR image, we propose a mid-tone optimization technique to calculate the modulation value. In particular, we use two conditions when deriving the modulation value: (1) the blacks are not clipped down to zero too aggressively; and (2) the number of codewords in the SDR image used to represent the mid-tone range of the HDR image is maximized. The modulation value can further be clipped to avoid over-shooting in both very dark images and very bright images. Temporal stabilization can also be used to smooth temporal variations of the luminosity in the SDR video. After the modulation value is determined, the SDR image can be obtained based on the HDR image and the modulation value.
1. A method for converting a first image to a second image, comprising:
determining a mid-tone level of the first image;
determining a modulation value responsive to a mapping function and the mid-tone level of the first image; and
converting the first image to the second image responsive to the modulation value and the mapping function, wherein the converting comprises scaling down the first image based on the modulation value, and wherein the scaled first image is converted to the second image based on the mapping function.
2. The method of claim 1 , wherein the mid-tone level is determined based on a black level and a white level.
3. The method of claim 2 , wherein the mid-tone level is determined as one of (1) a geometric mean and (2) a logarithm mean of the black level and the white level.
4. The method of claim 2 , wherein the black level and the white level are determined based on a histogram.
5. The method of claim 2 , wherein the modulation value is determined such that the conversion of the first image preserves information in a black area of the first image in the second image.
6. The method of claim 5 , wherein a codeword, in the second image, corresponding to the black level in the first image exceeds a threshold.
7. The method of claim 1 , wherein the first image is one of a plurality of images included in a video, and wherein the determining the modulation value is performed for each of the plurality of images, wherein the modulation values for the plurality of images are temporally smoothed.
8. The method of claim 1 , wherein the determining the modulation value substantially maximizes a number of codewords used in the second image to represent a range around the mid-tone level in the first image.
9. The method of claim 8 , wherein the determining the modulation value is responsive to a steepest slope of the mapping function.
10. The method of claim 1 , wherein the mapping function is formulated as g Ba (z)=M SDR f(z)/f(P/Ba), wherein P is an upper bound of a dynamic range for the first image, Ba is the modulation value, M SDR is an upper bound of a data range of the second image, and f( ) is a function.
11. The method of claim 10 , wherein the f( ) function is an Slog function.
12. An apparatus for converting a first image to a second image, comprising:
a communication interface configured to access the first image; and
one or more processors configured to
determine a mid-tone level of the first image,
determine a modulation value responsive to a mapping function and the mid-tone level of the first image, and
convert the first image to the second image responsive to the modulation value and the mapping function, wherein the processor is configured to scale down the first image based on the modulation value, and to convert the scaled first image to the second image based on the mapping function.
13. The apparatus of claim 12 , wherein the mid-tone level is determined based on a black level and a white level.
14. The apparatus of claim 12 , wherein the modulation value is determined to substantially maximize a number of codewords used in the second image to represent a range around the mid-tone level in the first image.
15. The apparatus of claim 13 , wherein the mid-tone level is determined as one of (1) a geometric mean and (2) a logarithm mean of the black level and the white level.
16. The apparatus of claim 13 , wherein the black level and the white level are determined based on a histogram.
17. The apparatus of claim 13 , wherein the modulation value is determined such that the conversion of the first image preserves information in a black area of the first image in the second image.
18. The apparatus of claim 17 , wherein a codeword, in the second image, corresponding to the black level in the first image exceeds a threshold.
19. The apparatus of claim 12 , wherein the first image is one of a plurality of images included in a video, and wherein the one or more processors are configured to determine the modulation value for each of the plurality of images, wherein the modulation values for the plurality of images are temporally smoothed.
20. A non-transitory computer readable storage medium having stored thereon instructions that, when executed, implement a method for converting a first image to a second image, the method comprising:
determining a mid-tone level of the first image;
determining a modulation value responsive to a mapping function and the mid-tone level of the first image; and
converting the first image to the second image responsive to the modulation value and the mapping function, wherein the converting comprises scaling down the first image based on the modulation value, and wherein the scaled first image is converted to the second image based on the mapping function.