Signal reshaping approximation

Patent metadata

Field	Value
Publication number	US-10080026-B2
Application number	US-201615546792-A
Country	US
Kind code	B2
Filing date	Mar 17, 2016
Priority date	Mar 20, 2015
Publication date	Sep 18, 2018
Grant date	Sep 18, 2018

Abstract

Statistical values are computed based on received source images. An adaptive reshaping function is selected for one or more source images based on the one or more statistical values. A portion of source video content is adaptively reshaped, based on the selected adaptive reshaping function to generate a portion of reshaped video content. The portion of source video content is represented by the one or more source images. An approximation of an inverse of the selected adaptive reshaping function is generated. The reshaped video content and a set of adaptive reshaping parameters defining the approximation of the inverse of the selected adaptive reshaping function are encoded into a reshaped video signal. The reshaped video signal may be processed by a downstream recipient device to generate a version of reconstructed source images, for example, for rendering with a display device.

First claim

What is claimed is:

1. A method, comprising:
computing one or more statistical values based on one or more source images in a sequence of source images;
selecting, based on the one or more statistical values, an adaptive reshaping function for the one or more source images, the adaptive reshaping function mapping source codewords to reshaped codewords;
adaptively reshaping, based at least in part on the selected adaptive reshaping function, a portion of source video content to generate a portion of reshaped video content, the portion of source video content being represented by the one or more source images;
generating an approximation of an inverse of the selected adaptive reshaping function, comprising:
determining a target lookup table (LUT) comprising key-value pairs representing the inverse of the selected adaptive reshaping function;
generating a first approximation of the target LUT by performing a forward search from small keys to large keys in the key-value pairs of the LUT;
generating a second approximation of the target LUT by performing a backward search from large keys to small keys in the key-value pairs of the LUT;
selecting one of the first approximation and the second approximation by comparing approximation errors respectively generated by the forward search and the backward search;

encoding the reshaped video content and a set of adaptive reshaping parameters that define the approximation of the inverse of the selected adaptive reshaping function into a reshaped video signal,
wherein the approximation of the inverse of the selected adaptive reshaping function is represented by a set of second order polynomials, the method further comprising:
determining a continuity condition for approximating the target LUT;
based on the continuity condition, selecting a first stopping rule for the forward search used to approximate the target LUT and a second stopping rule for the backward search used to approximate the target LUT;
generating the first approximation based at least in part on the first stopping rule; and
generating the second approximation based at least in part on the second stopping rule.

2. The method as recited in claim 1 , wherein the portion of the reshaped video content comprises one or more reshaped images.

3. The method as recited in claim 1 , wherein the one or more source images form a scene.

4. The method as recited in claim 1 , wherein the target LUT is an optimal backward LUT generated by averages of source codeword values that are mapped to each reshaped codeword value in a plurality of reshaped codeword values that are used to reshape the source video content.

5. The method as recited in claim 1 , wherein the one or more statistic values include at least one of a maximum value, a minimum value, a mean value, a median value, an average value, or a standard deviation value, as determined based on source codewords in the one or more source images.

6. The method as recited in claim 1 , wherein at least one of the selected adaptive reshaping function or the inverse of the selected adaptive reshaping function comprises one or more of analytical functions, non-analytical functions, lookup tables (LUTs), sigmoid functions, power functions, or piecewise functions.

7. The method as recited in claim 1 , wherein coefficients for polynomials in the set of polynomials are determined based on minimizing differences between values given by the polynomials and values given in a target lookup table (LUT) that represents the inverse of the selected adaptive reshaping function.

8. The method as recited in claim 1 , further comprising selecting a continuity condition for generating the set of polynomials based on a type of function determined for the inverse of the selected adaptive reshaping function.

9. The method as recited in claim 1 , wherein the set of polynomials are dynamically determined while the one or more source images are being processed for adaptive reshaping.

10. The method as recited in claim 1 , further comprising classifying the one or more source images as one of images comprising smooth bright areas, images comprising smooth dark areas, or mid-tone images.

11. The method of claim 1 , wherein the adaptive reshaping function is approximated using two or more second order polynomials and computing m p coefficients of the p-th polynomial comprises:
determining a first (beta) look-up table (LUT) based on a function of the reshaped values for pixel values in the sequence of source images;
determining a second (alpha) LUT based on a function of the original pixel values in a source image and the reshaped pixel values;
determining a B p matrix based on the first LUT;
determining an a p vector based on the second LUT; and
computing the m p coefficients of the p-th polynomial as B p −1 a p .

12. The method of claim 11 , wherein for a β[k,j] element of the first LUT:
β[ k,j]= 1, for k= 0≤ k ≤Max, j= 0

β[ k,j]=Σ i=0 k s i j , for 0≤ k ≤Max, 1≤ j≤ 4,

where Max denotes the maximum pixel value of the reshaped pixels s i corresponding to pixels ν i of a source image in the sequence of source images.

13. The method of claim 11 , wherein for an α[k,j] element of the second LUT, for 0≤j≤3
α[ k,j]=Σ i=0 k ν i s i j , for 0≤ k ≤Max,

where Max denotes the maximum pixel value of the reshaped pixels s i corresponding to pixels ν i of a source image in the sequence of source images.

14. The method of claim 1 , wherein generating a LUT for an inverse reshaping function comprises:
generating a histogram of reshaped values based on the forward reshaping function;
generating a cumulative table, wherein an entry in the cumulative table comprises the sum of original pixel values mapped to the same reshaped value; and
generating the LUT for the inverse reshaping function based on the histogram of the reshaped values and the cumulative table.

15. The method of claim 1 , wherein the adaptive reshaping function is approximated using two or more second order polynomials and pivot points for the two or more polynomials are selected according to an iterative method.

16. The method of claim 15 , wherein the iterative method further comprises:
setting an initial error threshold;
fitting segments of the adaptive reshaping function so that a fitting error of each of the one or more polynomials with a corresponding segment in the adaptive reshaping function does not exceed the initial error threshold;
determining a minimum of all the fitting errors across all segments of the adaptive reshaping function; and
repeating the fitting process for a new error threshold, wherein the new error threshold is smaller than the minimum of all the fitting errors.

17. The method of claim 16 , further comprising terminating the iterative method when the minimum of all fitting errors in the current iteration equals within a threshold the minimum of all fitting errors in the previous iteration.

18. In a decoder, a method for reconstructing video using a processor, the method comprising:
retrieving reshaped video content and a set of adaptive reshaping parameters defining a set of second order polynomials approximating an inverse of an adaptive reshaping function from a reshaped video signal, the inverse of the adaptive reshaping function mapping reshaped codewords to reconstructed source codewords;
the reshaped video content being generated by an upstream device based at least in part on the adaptive reshaping function, wherein the adaptive reshaping function is selected according to cl

Assignees

• Dolby Laboratories Licensing Corp

Inventors

• Su Guan-Ming
• Wang Qiuwei
• Chou Hsuan-Ting
• Kamballur Kottayil Navaneeth

Classifications

• H04N19/30 Primary

  using hierarchical techniques, e.g. scalability (H04N19/63 takes precedence) · CPC title

• H04N19/1887 Primary

  the unit being a variable length codeword · CPC title

• H04N21/47205

  for manipulating displayed content, e.g. interacting with MPEG-4 objects, editing locally · CPC title

• H04N19/70

  characterised by syntax aspects related to video coding, e.g. related to compression standards · CPC title

• H04N19/192

  the adaptation method, adaptation tool or adaptation type being iterative or recursive · CPC title