Inverse tone mapping for bit-depth scalable image coding adapted to variable block sizes

1. A method for inverse tone mapping of an image in a decoder, in which the image includes a plurality of blocks, comprising for each block of each color channel of the image the steps of:

determining a scaling factor for a current block by adding a predicted scaling factor for the current block to a difference between the predicted scaling factor and the scaling factor of an adjacent block;

determining an offset value for the current block by adding a predicted offset for the current block to a difference between the predicted offset value and the offset value of the adjacent block; and

applying the scaling factor and the offset value to pixel intensity values of the current block to produce a mapped block in which a bit-depth of the mapped block is greater than the bit-depth of the current block, in which the blocks have different sizes, wherein the steps of the method are performed in the decoder.

2. The method of claim 1 , in which the inverse tone mapping as adapted to the block having the different sizes.

3. The method of claim 2 , further comprising:

acquiring the image such that various regions the image have different lighting conditions, and the different sizes of the blocks are adapted to the various regions.

4. The method of claim 3 , in which a larger block size is used for region with uniform tone mapping, and a smaller block size is used regions with non-uniform tone mapping.

5. The method of claim 1 , in which the different sizes depend on a rate distortion cost (RDC) of encoding the blocks.

6. The method of claim 5 , in which the RDC is D+λR, where D is distortion measured in MSE, R is a rate is a total number of bits used to encode a particular block, including overhead bits for encoding the scaling factor and the offset value, and λ is a Lagrangian multiplier.

7. The method of claim 5 , further comprising:

determining the RDC for the particular block as RDC(M×N), where M and N are dimensions of the particular block in terms of pixels; and

partitioning the particular block into sub-blocks as RDC(M/2×N/2);

determining a sum of the RDCs for the sub-blocks blocks as RDC(M/2×N/2), and if RDC(M,N)≦RDC(M/2,N/2), then retaining the size of the particular block as M×N, and otherwise iterating the partitioning and determining of the sum until RCD(M/k, N/k)≦RCD(M/k/2, N/k/2), where k is a non-negative integer and increased by 1 for each iteration.

8. The method of claim 1 , in which the different sizes depend on a rate distortion cost (RDC) of encoding the blocks.

9. The method of claim 1 , in which the different sizes depend on the scaling factors of the blocks.

10. The method of claim 9 , in which a maximum block size is M×N and a minimum block size is m×n, and further comprising:

partitioning a particular block M×N block into m×n sub-blocks;

determining an optimal scaling factor for each sub-block; and

merging the sub-blocks having identical optimal scaling factors.

11. The method of claim 1 , in which the determining and applying steps are performed for a sequence of images.