Granted Patent B2

US 12,555,209 · App. 18/426,063 · Granted Feb 17, 2026

Inventors: Zeeshan Nadir (Allen, TX); Hamid Rahim Sheikh (Allen, TX); John William Glotzbach (Allen, TX)

Assignee: Samsung Electronics Co., Ltd.

G06T5/92G06T3/40G06T5/40G06T5/50G06T11/001G06V10/54G06T2207/20072G06T2207/20221

Abstract

Claims (88)

1 . An electronic device comprising:

at least one processing device configured to:

downsample an input frame;

perform global contrast enhancement of the input frame based on the downsampled input frame to generate a globally contrast-enhanced frame;

generate texture maps at different scale resolutions based on the downsampled input frame;

generate clipped histograms at the different scale resolutions based on the downsampled input frame;

generate lookup tables (LUTs) at the different scale resolutions based on the texture maps and the clipped histograms;

apply the LUTs at the different scale resolutions to the input frame in order to generate locally contrast-enhanced frames; and

blend the locally contrast-enhanced frames and the globally contrast-enhanced frame to generate an output frame.

2 . The electronic device of claim 1 , wherein, to perform the global contrast enhancement, the at least one processing device is configured to:

generate a global histogram using the downsampled input frame; and

perform global contrast enhancement of the input frame using the global histogram.

3 . The electronic device of claim 1 , wherein, to generate the clipped histograms, the at least one processing device is configured to:

generate histograms at different window sizes for each of the different scale resolutions;

determine, for each histogram, a clip limit for limiting contrast enhancement; and

clip each of the histograms using the clip limit corresponding to the histogram.

4 . The electronic device of claim 3 , wherein, to determine the clip limit, the at least one processing device is configured to:

identify, for each tile in the input frame, texture values in neighboring tiles; and

determine the clip limit based on a lowest of the texture values in the neighboring tiles.

5 . The electronic device of claim 3 , wherein, to determine the clip limit, the at least one processing device is configured to:

identify, for each tile in the input frame, texture values in neighboring tiles; and

determine the clip limit based on an average of the texture values in the neighboring tiles.

6 . The electronic device of claim 1 , wherein, to generate the texture maps, the at least one processing device is configured to:

determine a first local mean and a first local variance of a first window in the input frame;

determine a second local mean and a second local variance of a second window in the input frame that is partially shifted from the first window; and

generate the texture maps based on the first local variance and the second local variance.

7 . The electronic device of claim 6 , wherein, to determine the second local mean and the second local variance of the second window, the at least one processing device is configured to:

identify an overlap area of the first window and the second window;

identify a first unique area of the first window that includes an area of the downsampled input frame not included in the second window;

identify a second unique area of the second window that includes an area of the downsampled input frame not included in the first window; and

determine the second local mean and the second local variance without using pixel values in the first unique area.

8 . A non-transitory machine readable medium containing instructions that when executed cause at least one processor to:

downsample an input frame;

perform global contrast enhancement of the input frame based on the downsampled input frame to generate a globally contrast-enhanced frame;

generate texture maps at different scale resolutions based on the downsampled input frame;

generate clipped histograms at the different scale resolutions based on the downsampled input frame;

generate LUTs at the different scale resolutions based on the texture maps and the clipped histograms;

apply the LUTs at the different scale resolutions to the input frame in order to generate locally contrast-enhanced frames; and

blend the locally contrast-enhanced frames and the globally contrast-enhanced frame to generate an output frame.

9 . The non-transitory machine readable medium of claim 8 , wherein the instructions that when executed cause the at least one processor to generate the clipped histograms comprise instructions that when executed cause the at least one processor to:

generate histograms at different window sizes for each of the different scale resolutions;

determine, for each histogram, a clip limit for limiting contrast enhancement; and

clip each of the histograms using the clip limit corresponding to the histogram.

10 . The non-transitory machine readable medium of claim 9 , wherein the instructions that when executed cause the at least one processor to determine the clip limit comprise instructions that when executed cause the at least one processor to:

identify, for each tile in the input frame, texture values in neighboring tiles; and

determine the clip limit based on a lowest of the texture values in the neighboring tiles.

11 . The non-transitory machine readable medium of claim 9 , wherein the instructions that when executed cause the at least one processor to determine the clip limit comprise instructions that when executed cause the at least one processor to:

identify, for each tile in the input frame, texture values in neighboring tiles; and

determine the clip limit based on an average of the texture values in the neighboring tiles.

12 . The non-transitory machine readable medium of claim 8 , wherein the instructions that when executed cause the at least one processor to generate the texture maps comprise instructions that when executed cause the at least one processor to:

determine a first local mean and a first local variance of a first window in the input frame;

determine a second local mean and a second local variance of a second window in the input frame that is partially shifted from the first window; and

generate the texture maps based on the first local variance and the second local variance.

13 . The non-transitory machine readable medium of claim 12 , wherein the instructions that when executed cause the at least one processor to determine the second local mean and the second local variance of the second window comprise instructions that when executed cause the at least one processor to:

identify an overlap area of the first window and the second window;

identify a first unique area of the first window that includes an area of the downsampled input frame not included in the second window;

identify a second unique area of the second window that includes an area of the downsampled input frame not included in the first window; and

determine the second local mean and the second local variance without using pixel values in the first unique area.

14 . A method comprising:

downsampling an input frame;

performing global contrast enhancement of the input frame based on the downsampled input frame to generate a globally contrast-enhanced frame;

generating texture maps at different scale resolutions based on the downsampled input frame;

generating clipped histograms at the different scale resolutions based on the downsampled input frame;

generating lookup tables (LUTs) at the different scale resolutions based on the texture maps and the clipped histograms;

applying the LUTs at the different scale resolutions to the input frame in order to generate locally contrast-enhanced frames; and

blending the locally contrast-enhanced frames and the globally contrast-enhanced frame to generate an output frame.

15 . The method of claim 14 , wherein performing the global contrast enhancement comprises:

generating a global histogram using the downsampled input frame; and

performing global contrast enhancement of the input frame using the global histogram.

16 . The method of claim 14 , wherein generating the clipped histograms comprises:

generating histograms at different window sizes for each of the different scale resolutions;

determining, for each histogram, a clip limit for limiting contrast enhancement; and

clipping each of the histograms using the clip limit corresponding to the histogram.

17 . The method of claim 16 , wherein determining the clip limit comprises:

identifying, for each tile in the input frame, texture values in neighboring tiles; and

determining the clip limit based on a lowest of the texture values in the neighboring tiles.

18 . The method of claim 16 , wherein determining the clip limit comprises:

identifying, for each tile in the input frame, texture values in neighboring tiles; and

determining the clip limit based on an average of the texture values in the neighboring tiles.

19 . The method of claim 14 , wherein generating the texture maps comprises:

determining a first local mean and a first local variance of a first window in the input frame;

determining a second local mean and a second local variance of a second window in the input frame that is partially shifted from the first window; and

generating the texture maps based on the first local variance and the second local variance.

20 . The method of claim 19 , wherein determining the second local mean and the second local variance of the second window comprises:

identifying an overlap area of the first window and the second window;

identifying a first unique area of the first window that includes an area of the downsampled input frame not included in the second window;

identifying a second unique area of the second window that includes an area of the downsampled input frame not included in the first window; and

determining the second local mean and the second local variance without using pixel values in the first unique area.

Assignments (1)

Continuity (2)

References Cited (26)