Method of presenting wide dynamic range images and a system employing same

Patent metadata

Field	Value
Publication number	US-10922796-B2
Application number	US-201716317438-A
Country	US
Kind code	B2
Filing date	Jul 11, 2017
Priority date	Jul 11, 2016
Publication date	Feb 16, 2021
Grant date	Feb 16, 2021

Abstract

A method of converting a wide dynamic range (WDR) image to a low dynamic range (LDR) image have the steps of: (i) obtaining a transfer function that has a plurality of sub-functions, each sub-function corresponding to a non-overlapped input interval of the dynamic range of the WDR image; (ii) determining the intensity of each pixel of the LDR image by using the transfer function and at least the intensity value of the corresponding pixel of the WDR image; and (iii) outputting the LDR image.

First claim

What is claimed is:

1. A method of converting an input wide dynamic range (WDR) image into an output low dynamic range (LDR) image, the pixels of the input WDR image having a first dynamic range R WDR and the pixels of the output LDR image having a second dynamic range R LDR smaller than the first dynamic range, the method comprising:
representing each intensity value x within the first range R WDR as x=m×r s with m being a mantissa of x, r being a radix, s being an exponent of x, and × representing multiplication;
partitioning the first range R WDR into a plurality of input intervals X i with i being an integer and i≥0, based at least on values of the exponents within the first range R WDR , the input intervals X i being non-overlapped and spanning the first dynamic range R WDR ;
obtaining a transfer function ƒ(x) over the first dynamic range R WDR , the transfer function ƒ(x) comprising a plurality of sub-functions ƒ i (x), each sub-function ƒ i (x) being determined over one of the input intervals X i of the first dynamic range R WDR ;
determining the intensity y(p) of each pixel p of the output LDR image by using the transfer function ƒ(x) and at least the intensity value x(p) of the corresponding pixel p of the input WDR image; and
outputting the output LDR image.

2. The method of claim 1 wherein said obtaining the transfer function ƒ(x) comprises:
maintaining a lookup table of the transfer function ƒ(x), wherein said lookup table comprises a plurality of entries, each entry comprises a value of x represented as its mantissa m and exponent s, and the corresponding value of ƒ(x); and
wherein said determining the intensity y(p) of each pixel p of the output LDR image comprises:
representing the intensity value x(p) of the pixel p of the input WDR image in the mantissa-exponent form:
x ( p )= m ( p )× r s(p) ;

searching the lookup table using s(p) and m(p) for determining the corresponding value of ƒ(x(p)); and
using the determined value of ƒ(x(p)) as the intensity value of the pixel p of the output LDR image.

3. The method of claim 1 wherein the transfer function ƒ(x) is:
ƒ( x )=(log 2 ( a s ×x+b s )× c s )×log 2 ( g ( m,s ))× d p +k p ,

where a s , b s , c s , d p , and k p are parameters, and the function g(m,s) is a function of the mantissa m and the exponent s.

4. The method of claim 1 wherein said non-overlapped input intervals of the first dynamic range R WDR are partitioned based on at least the exponent s and a weighted combination D of a reference pixel intensity distribution and an input pixel intensity distribution:
D=α×D ref +β×D hist ,
where α and β are weight factors and α+β=1; D ref is the reference pixel intensity distribution; and D hist is the input pixel intensity distribution.

5. The method of claim 1 further comprising:
establishing a plurality of WDR image classes; and
establishing a plurality of transfer functions each for one of the plurality of WDR image classes;
wherein said obtaining the transfer function ƒ(x) comprises:
determining one of the plurality of WDR image classes for the input WDR image; and
determining the transfer function ƒ(x) as one of the plurality of transfer functions based on the determined WDR image class.

6. A system for converting an input WDR image into an output LDR image, the pixels of the input WDR image having a first dynamic range R WDR and the pixels of the output LDR image having a second dynamic range R LDR smaller than the first dynamic range, the system comprising:
an input for inputting the input WDR image;
an output for outputting the output LDR image; and
a processing structure comprising one or more computing processors or one or more circuits, the processing structure functionally coupled to the input and the output, the processing structure being configured for:
representing each intensity value x within the first range R WDR as x=m×r s with m being a mantissa of x, r being a radix, s being an exponent of x, and × representing multiplication;
partitioning the first range R WDR into a plurality of input intervals X i with i being an integer and i≥0, based at least on values of the exponents within the first range R WDR , the input intervals X i being non-overlapped and spanning the first dynamic range R WDR ,
obtaining a transfer function ƒ(x) over the first dynamic range R WDR , the transfer function ƒ(x) comprising a plurality of sub-functions ƒ i (x), each sub-function ƒ i (x) being determined over one of the input intervals X i of the first dynamic range R WDR ,
determining the intensity y(p) of each pixel p of the output LDR image by using the transfer function ƒ(x) and at least the intensity value x(p) of the corresponding pixel p of the input WDR image; and
outputting the output LDR image.

7. The system of claim 6 wherein said obtaining the transfer function ƒ(x) comprises:
maintaining a lookup table of the transfer function ƒ(x), wherein said lookup table comprises a plurality of entries, each entry comprises a value of x represented as its mantissa m and exponent s, and the corresponding value of ƒ(x); and
wherein said determining the intensity y(p) of each pixel p of the output LDR image comprises:
representing the intensity value x(p) of the pixel p of the input WDR image in the mantissa-exponent form:
x ( p )= m ( p )× r s(p) ;

searching the lookup table using s(p) and m(p) for determining the corresponding value of ƒ(x(p)); and
using the determined value of ƒ(x(p)) as the intensity value of the pixel p of the output LDR image.

8. The system of claim 6 wherein the transfer function ƒ(x) is:
ƒ( x )=(log 2 ( a s ×x+b s )+ c s )×log 2 ( g ( m,s ))× d p +k p ,

where a s , b s , c s , d p , and k p are parameters, and the function g(m,s) is a function of mantissa m and exponent s.

9. The system of claim 6 wherein r=2, and m and s are integers greater than or equal to zero.

10. The system of claim 6 wherein said non-overlapped input intervals of the first dynamic range R WDR are partitioned based on at least the exponent s and a weighted combination D of a reference pixel intensity distribution and an input pixel intensity distribution:
D=α×D ref +β×D hist ,
where α and β are weight factors and α+β=1; D ref is the reference pixel intensity distribution; and D hist is the input pixel intensity distribution.

11. The system of claim 10 wherein r=2, and m and s are integers greater than or equal to zero; wherein D=D(s), D ref =D ref (S), D hist =D hist (s) are functions of s; wherein the first pixel intensity range R WDR is between zero and a maximum value m MAX , and the second pixel intensity range R LDR is between zero and a maximum value y MAX ; wherein the transfer function ƒ(x) is:

f
⁡

(
x
)

=

(

log
2

⁡

(

x
+

Assignees

• Uti Lp
• Tonetech Inc

Inventors

• Yadid-Pecht Orly
• Yang Jie
• Hore Alain
• Shahnovich Ulian

Classifications

• H04N23/741

  by increasing the dynamic range of the image compared to the dynamic range of the electronic image sensors · CPC title

• H04N23/80

  Camera processing pipelines; Components thereof · CPC title

• G06T5/94

  based on local image properties, e.g. for local contrast enhancement · CPC title

• H04N5/20 Primary

  Circuitry for controlling amplitude response · CPC title

• G06T7/10

  Segmentation; Edge detection (motion-based segmentation G06T7/215) · CPC title