Global approximation to spatially varying tone mapping operators
Techniques to generate global tone-mapping operators (G-TMOs) that, when applied to high dynamic range images, visually approximate the use of spatially varying tone-mapping operators (SV-TMOs) are described. The disclosed G-TMOs provide substantially the same visual benefits as SV-TMOs but do not suffer from spatial artifacts such as halos and are, in addition, computationally efficient compared to SV-TMOs. In general, G-TMOs may be identified based on application of a SV-TMO to a down-sampled version of a full-resolution input image (e.g., a thumbnail). An optimized mapping between the SV-TMO's input and output constitutes the G-TMO. It has been unexpectedly discovered that when optimized (e.g., to minimize the error between the SV-TMO's input and output), G-TMOs so generated provide an excellent visual approximation to the SV-TMO (as applied to the full-resolution image).
1. A non-transitory program storage device comprising instructions stored thereon to cause one or more processors to:
obtain a grayscale version of a high dynamic range (HDR) color input image;
downsample the grayscale version of the HDR color input image to generate a down-sampled grayscale input image;
apply a spatially varying tone mapping operator (SV-TMO) to the down-sampled grayscale input image to generate a down-sampled grayscale output image;
determine a global tone mapping operator (G-TMO) according to, at least in part, a Pool-Adjacent-Violators-Algorithm (PAVA); and
apply the G-TMO to the grayscale version of the HDR color input image to generate a low dynamic range (LDR) grayscale output image.
2. The non-transitory program storage device of claim 1 , wherein the instructions to determine the G-TMO further comprises instructions to cause the one or more processors to:
determine a first G-TMO according to the PAVA; and
apply a smoothing filter to the first G-TMO to generate the G-TMO.
3. The non-transitory program storage device of claim 1 , wherein the instructions to determine the G-TMO comprise instructions to cause the one or more processors to determine a mapping between each pixel in the down-sampled grayscale input image to a corresponding pixel in the down-sampled grayscale output image, wherein the mapping is selected to minimize a specified error criterion.
4. The non-transitory program storage device of claim 3 , wherein the specified error criterion comprises a root means squares error criteria.
5. The non-transitory program storage device of claim 1 , further comprising instructions to cause the one or more processors to convert the LDR grayscale output image to a LDR color output image.
6. The non-transitory program storage device of claim 2 , wherein each pixel in the color output image is based, in part, on a ratio of the corresponding pixels in the grayscale output image and the grayscale version of the color input image.
7. The non-transitory program storage device of claim 1 , wherein the instructions to obtain the grayscale version of the HDR color input image comprise instructions to cause the one or more processors to obtain a brightness channel of the HDR color input image, wherein the HDR color input image comprises a brightness channel and one or more chrominance channels.
8. The non-transitory program storage device of claim 1 , wherein the instructions to generate the down-sampled grayscale input image comprise instructions to cause the one or more processors to generate a thumbnail of the grayscale version of the HDR color input image.
9. The non-transitory program storage device of claim 1 , wherein the instructions to determine the first G-TMO according to the PAVA comprise instructions to cause the one or more processors to quantize values from the down-sampled grayscale input image into a specified number of levels, wherein the specified number of levels is less than the number of possible brightness levels that a pixel in the down-sampled grayscale input image may have.
10. The non-transitory program storage device of claim 1 , wherein the instructions to generate the down-sampled grayscale output image comprise instructions to cause the one or more processors to:
apply the G-TMO to the grayscale version of the HDR color input image to generate a first grayscale image; and
perform detail recovery operations on the first grayscale image to generate a grayscale output image.
11. The non-transitory program storage device of claim 10 , wherein the instructions to perform the detail recovery operations comprise instructions to cause the one or more processors to apply a bilateral filter to a combination of the grayscale version of the HDR color input image and the first grayscale image.
12. The non-transitory program storage device of claim 10 , wherein the instructions to generate the grayscale output image comprise instructions to cause the one or more processors to:
perform detail recovery operations on the first grayscale image to generate a second grayscale image; and
apply an unsharp mask to the second grayscale image to generate a grayscale output image.
13. An electronic device, comprising:
a display element;
a memory operatively coupled to the display element; and
one or more processing units operatively coupled to the display element and the memory, and adapted to execute instructions stored in the memory to:
obtain a grayscale version of a high dynamic range (HDR) color input image;
downsample the grayscale version of the HDR color input image to generate a down-sampled grayscale input image;
apply a spatially varying tone mapping operator (SV-TMO) to the down-sampled grayscale input image to generate a down-sampled grayscale output image;
determine a global tone mapping operator (G-TMO) according to, at least in part, a Pool-Adjacent-Violators-Algorithm (PAVA); and
apply the G-TMO to the grayscale version of the HDR color input image to generate a low dynamic range (LDR) grayscale output image.
14. The electronic device of claim 13 , wherein the one or more processing units are adapted to execute instructions to:
determine a first G-TMO according to the PAVA; and
apply a smoothing filter to the first G-TMO to generate the G-TMO.
15. The electronic device of claim 13 , wherein the one or more processing units are adapted to execute instructions to determine a mapping between each pixel in the down-sampled grayscale input image to a corresponding pixel in the down-sampled grayscale output image, wherein the mapping is selected to minimize a specified error criterion.
16. The electronic device of claim 15 , wherein the specified error criterion comprises a root means squares error criteria.
17. The electronic device of claim 13 , wherein the one or more processing units are adapted to execute instructions to convert the LDR grayscale output image to a LDR color output image.
18. The electronic device of claim 17 , wherein each pixel in the color output image is based, in part, on a ratio of the corresponding pixels in the grayscale output image and the grayscale version of the color input image.
19. The electronic device of claim 13 , wherein the one or more processing units are adapted to execute instructions to obtain a brightness channel of the HDR color input image, wherein the HDR color input image comprises a brightness channel and one or more chrominance channels.
20. The electronic device of claim 13 , wherein the one or more processing units are adapted to execute instructions to generate a thumbnail of the grayscale version of the HDR color input image.
21. The electronic device of claim 13 , wherein the one or more processing units are adapted to execute instructions to quantize values from the down-sampled grayscale input image into a specified number of levels, wherein the specified number of levels is less than the number of possible brightness levels that a pixel in the down-sampled grayscale input image may have.
22. The electronic device of claim 13 , wherein the one or more processing units are adapted to execute instructions to:
apply the G-TMO to the grayscale version of the HDR color input image to generate a first grayscale image; and
perform detail recovery operations on the first grayscale image to generate a grayscale output image.
23. The electronic device of claim 22 , wherein the one or more processing units are adapted to execute instructions to apply a bilateral filter to a combination of the grayscale version of the HDR color input image and the first grayscale image.
24. The electronic device of claim 22 , wherein the one or more processing units are adapted to execute instructions to
perform detail recovery operations on the first grayscale image to generate a second grayscale image; and
apply an unsharp mask to the second grayscale image to generate a grayscale output image.
25. An image conversion method, comprising:
obtaining a grayscale version of a high dynamic range (HDR) color input image;
downsampling the grayscale version of the HDR color input image to generate a down-sampled grayscale input image;
applying a spatially varying tone mapping operator (SV-TMO) to the down-sampled grayscale input image to generate a down-sampled grayscale output image;
determining a global tone mapping operator (G-TMO) according to, at least in part, a Pool-Adjacent-Violators-Algorithm (PAVA); and
applying the G-TMO to the grayscale version of the HDR color input image to generate a low dynamic range (LDR) grayscale output image.