Apparatus and methods for encoding, decoding and representing high dynamic range images
A data structure defining a high dynamic range image comprises a tone map having a reduced dynamic range and HDR information. The high dynamic range image can be reconstructed from the tone map and the HDR information. The data structure can be backwards compatible with legacy hardware or software viewers. The data structure may comprise a JFIF file having the tone map encoded as a JPEG image with the HDR information in an application extension or comment field of the JFIF file, or a MPEG file having the tone map encoded as a MPEG image with the HDR information in a video or audio channel of the MPEG file. Apparatus and methods for encoding or decoding the data structure may apply pre- or post correction to compensate for lossy encoding of the high dynamic range information.
1. A method for encoding a high dynamic range (HDR) image, the method comprising:
generating a lower dynamic range (LDR) image by tone mapping the HDR image, the LDR image having a lower dynamic range than the HDR image;
computing an original ratio image, wherein each pixel value of the original ratio image comprises a ratio of a luminance of a pixel of the HDR image and a luminance of a corresponding pixel of the LDR image; and
computing a transformed ratio image by applying a non-linear invertible function to the original ratio image;
generating a compressed transformed ratio image by compressing the transformed ratio image; and
generating a compressed file that includes both the LDR image and the compressed transformed ratio image.
2. The method of claim 1 , wherein the tone mapping clips luminance values of the HDR image.
3. The method of claim 1 , wherein the tone mapping does not clip luminance values of the HDR image.
4. The method of claim 1 , wherein the compressed file is a JPEG compressed file.
5. The method of claim 4 , wherein the ratio image is separated as an application extension from the LDR image in the JPEG compressed file.
6. The method of claim 1 , wherein the ratio image is ASCII text in the compressed file.
7. The method of claim 1 , wherein the compressed file further includes a flag indicative of HDR processing.
8. The method of claim 1 , wherein the generating of the ratio image is performed by an encoder that includes a processor.
9. A non-transitory computer-readable storage medium encoded with instructions for causing a processor to perform the method of claim 1 .
10. A method for decoding a high dynamic range (HDR) image, the method comprising:
receiving a compressed file, the compressed file including a lower dynamic range (LDR) image, wherein the LDR image has a lower dynamic range than a HDR image;
receiving a compressed transformed ratio image;
generating a transformed ratio image by decompressing the compressed transformed ratio image;
applying an inverse function of a non-linear invertible function to the transformed ratio image to generate a decoded ratio image, wherein each pixel value of the decoded ratio image is a ratio of a luminance of a pixel of the HDR image and a luminance of a corresponding pixel of the LDR image; and
reconstructing a reconstructed HDR image, the reconstructed HDR image determined by the LDR image and the decoded ratio image.
11. The method of claim 10 , wherein the LDR image is a tone map corresponding to the HDR image.
12. The method of claim 11 , wherein tone map includes a clipped luminance value of the HDR image.
13. The method of claim 10 , wherein the decoded ratio image is ASCII text.
14. The method of claim 10 , further comprising rendering the reconstructed HDR image on a display.
15. The method of claim 10 , wherein the compressed file further includes a flag indicative of HDR processing.
16. The method of claim 10 , wherein the reconstructing is performed by a decoder that includes a processor.
17. A non-transitory computer-readable storage medium encoded with instructions for causing a processor to perform the method of claim 10 .
18. The method of claim 1 , wherein the non-linear invertible function is a logarithmic function or a square-root function.
19. The method of claim 1 , wherein each pixel of the transformed ratio image is represented by one byte (8 bits).
20. The method of claim 10 , wherein the second function is an inverse logarithm function or an inverse square root function.
21. The method of claim 10 , further comprising the steps of:
applying a post-correction to the decoded ratio image prior to reconstructing the reconstructed HDR image, wherein the post-correction comprises calculating a corrected ratio image based on a function of the decoded ratio image, a set of luminance values of tone map data generated from the HDR image, and a set of downsampled luminance values of tone map data generated from the HDR image.
22. The method of claim 21 , wherein calculating the corrected ratio image comprises calculating:
RI
CORRECTED
=
RI
×
(
L
(
TM
)
L
(
TM
R
)
)
σ
or a mathematical equivalent thereof, where RI CORRECTED is the corrected ratio image, RI is the decoded ratio image, L(TM) is the set of luminance values of tone map data generated from the HDR image, L(TM R ) is the set of downsampled luminance values of tone map data generated from the HDR image, and σ is a measure of variance in the ratio of L(TM) to L(TM R ).