Adaptive reshaping for layered coding of enhanced dynamic range signals

Patent metadata

Field	Value
Publication number	US-9584811-B2
Application number	US-201414898574-A
Country	US
Kind code	B2
Filing date	Jun 16, 2014
Priority date	Jun 17, 2013
Publication date	Feb 28, 2017
Grant date	Feb 28, 2017

Abstract

An encoder receives an input enhanced dynamic range (EDR) image to be coded in a layered representation. Input images may be gamma-coded or perceptually-coded using a bit-depth format not supported by one or more video encoders. The input image is remapped to one or more quantized layers to generate output code words suitable for compression using the available video encoders. Algorithms to determine optimum function parameters for linear and non-linear mapping functions are presented. Given a mapping function, the reverse mapping function may be transmitted to a decoder as a look-up table or it may be approximated using a piecewise polynomial approximation. A polynomial approximation technique for representing reverse-mapping functions and chromaticity translation schemes to reduce color shifts are also presented.

First claim

What is claimed is:

1. A method for coding a sequence of input enhanced dynamic range (EDR) images comprising image blocks, the method comprising:
receiving said sequence of input enhanced dynamic range (EDR) images;
computing block-complexity metrics for one or more of the image blocks in at least one input image in the sequence of input EDR images, said block-complexity metrics being representative for a variation in pixel values;
constructing a first set of the image blocks, the first set comprising image blocks for which their computed block-complexity metric satisfies a predetermined criterion;
for each block in the first set of the image blocks determining an optimal slope (k(j,n)) according to a slope generation function for a linear quantization model, said optimal slope representing a minimal slope of the linear quantization model for said block;
for each input code word (v c ) of the sequence of input EDR images:
constructing a second set of image blocks, the second set comprising blocks that belong to the first set of the image blocks and wherein the input code word is within minimum and maximum pixel values of the image block; and
generating an envelope slope (k(v c )) for the input code word, said envelope slope being computed using the maximum optimal slope among the optimal slopes of the blocks in the second set of image blocks;

computing a sum (k) of all the envelope slopes for all code words; and
for each input code word:
generating a cumulative slope (K(v c )), the cumulative slope comprising a sum of the envelope slopes up to and including said input code word; and
generating a mapping function between the input code word and an output code word, the mapping function computed from the cumulative slope of the code word and the sum of all the envelope slopes for all code words.

2. The method of claim 1 , further comprising:
applying the mapping function to the input EDR images to generate reshaped images;
decomposing the reshaped images into one or more layers; and
encoding the one or more layers using one or more video encoders.

3. The method of claim 2 , wherein, given a reshaped pixel s, and L layers, the decomposing step comprises mapping the s pixel into s l pixel values
s l =Clip3( s,p l ,p l+1 −1)− p l ,

wherein l=0, 1, . . . , L−1, denotes one of the L layers, Clip3( ) is a clipping function that clips the reshaped pixel s between the values of p l and p l+1 −1, and p l denotes the smallest pixel value of the reshaped sequence at level l.

4. The method of claim 3 , wherein p 0 =0 and

p
i

=

∑

j
=
0

i
-
1

⁢

⁢

N
j

⁢

⁢
for
⁢

⁢
i

=
1

,
…
⁢

,
L
,

where N j denotes the number of code words available for a video encoder at level j.

5. The method of claim 1 , wherein the linear quantization model comprises a function denoted by

s
i

=

round
⁡

(

k
⁡

(

j
,
n

)

⁢

(

c
H

-

c
L

)

⁢

(

v
i

-

v
L

v
H

-

v
L

)

+

c
L

)

,

where v L and v H denote a minimum and a maximum code word value in the EDR input sequence, c L and c H denote a minimum and a maximum output code value, k(j,n) denotes a quantization slope for the n-th block in the j-th frame in the EDR input sequence, v i denotes an input code word and s i denotes a corresponding output code word.

6. The method of claim 5 , wherein the optimal slope k(j,n) is generated by computing

k
⁡

(

j
,
n

)

=

T
th

⁢

v
H

-

v
L

Assignees

• Dolby Laboratories Licensing Corp

Inventors

• Su Guan-Ming

Classifications

• H04N19/184

  the unit being bits, e.g. of the compressed video stream · CPC title

• H04N19/85

  using pre-processing or post-processing specially adapted for video compression · CPC title

• H04N19/176

  the region being a block, e.g. a macroblock · CPC title

• H04N19/136

  Incoming video signal characteristics or properties · CPC title

• H04N19/36

  Scalability techniques involving formatting the layers as a function of picture distortion after decoding, e.g. signal-to-noise [SNR] scalability · CPC title