IP Library › Granted Patent US 9,865,035
Granted Patent B2
US 9,865,035 · App. 14/843,974 · Granted Jan 9, 2018

Image scaling techniques

Inventor: Walter E. Donovan (Saratoga, CA)
Assignee: NVIDIA CORPORATION
G06T3/4007
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 9,865,035
App. No.
14/843,974
Granted
Jan 9, 2018
Kind
B2
Abstract

Image scaling techniques, in accordance with embodiments of the present technology, include directionally interpolating blocks of pixel data of an image, sharpening the directional interpolated blocks of pixel data, and optionally clamping the sharpened, directional interpolated blocks of pixel data.

Claims (193)

1. A method comprising

receiving a plurality of blocks of pixel data of an image;

directionally interpolating the plurality of blocks of pixel data including

calculating metrics M45 and M135 as the sum of absolute difference of the luminances

M

45

=

∑

45

°

⁢

⁢

pairs

⁢

⁢

i

,

j

⁢

lum

i

-

lum

j

M

135

=

∑

135

°

⁢

⁢

pairs

⁢

⁢

i

,

j

⁢

lum

i

-

lum

j

of diagonally-adjacent pixels for each block of pixel data;

color sharpening the directional interpolated blocks of pixel data;

iteratively repeating the combination of the directionally interpolating and color sharpening until a predetermined scaling has been achieved; and

outputting the color sharpened directional interpolated blocks of pixel data as a scaled image when the predetermined scaling has been achieved.

2. The method according to claim 1 , wherein the direction of interpolation is chosen based upon the metrics M 45 and M 135 .

3. The method according to claim 2 , wherein directionally interpolating includes applying a linear filter when there metrics M 45 and M 135 fails to select the direction of interpolation.

4. The method according to claim 1 , wherein color sharpening the directional interpolated blocks of pixel data includes applying a low pass filter.

5. A method comprising:

receiving a plurality of blocks of pixel data of an image;

directionally interpolating the plurality of blocks of pixel data including

calculating metrics M 45 and M 135 as the sum of absolute difference of the luminaces

M

45

=

∑

45

°

⁢

⁢

pairs

⁢

⁢

i

,

j

⁢

lum

i

-

lum

j

M

135

=

∑

135

°

⁢

⁢

pairs

⁢

⁢

i

,

j

⁢

lum

i

-

lum

j

of diagonal pixels for each blocks of pixel data;

sharpening the directional interpolated blocks of pixel data;

clamping the sharpened, directional interpolated blocks of pixel data; and

outputting the clamped, sharpened, directional interpolated blocks of pixel data as a scaled image.

6. The method of according to claim 5 , wherein directionally interpolating the plurality of blocks of pixel data includes applying a directional non-linear filter to the blocks of pixel data.

7. The method according to claim 5 , wherein the direction of the interpolation is chosen based upon the metrics M 45 and M 135 , and a linear filter is applied when the metrics M 45 and M 135 fail to select the direction of interpolation.

8. The method according to claim 5 , wherein sharpening the directional interpolated blocks of pixel data includes applying a sharpening filter to the directional interpolated blocks of pixel data.

9. The method according to claim 5 , wherein sharpening the directional interpolated blocks of pixel data includes applying a lowpass filter.

10. The method according to claim 5 , wherein:

sharpening the directional interpolated blocks of pixel data comprises color sharpening of the directional interpolated blocks of pixel data; and

clamping the sharpened, directional interpolated blocks of pixel data comprises clamping the sharpened, directional, interpolated blocks of pixel data to its representable range.

11. The method according to claim 5 , wherein:

sharpening the directional interpolated blocks of pixel data comprises luminance sharpening the directional interpolated blocks of pixel data; and

clamping the sharpened directional interpolated blocks of pixel data comprises clamping the sharpened, directional interpolated blocks of pixel data to a previous luminance value in a punctuated neighborhood of each respective block of pixel data.

12. A method comprising: receiving blocks of pixel data on an image;

receiving blocks of pixel data of an image;

receiving an image scaling factor;

directionally interpolating the blocks of pixel data, if the image scaling factor is greater than 1.0 and less than 2.0:

downsampling the directional interpolated blocks of pixel data, if the image scaling factor is greater than 1.0 and less than 2.0;

color sharpening the down sampled, directional interpolated blocks of pixel data, if the image scaling factor is greater than 1.0 and less than 2.0;

luminance clamping the color sharpened, downsampled, directional interpolated blocks of pixel data, if he image scaling factor is greater than 1.0 and less than 2.0;

outputting the luminance clamped, color sharpened, down sampled, directional interpolated blocks of pixel data as a scaled image, if the image scaling factor is greater than 1.0 and less than 2.0.

13. The method according to claim 12 , wherein directionally interpolating the blocks of pixel data includes applying a directional non-linear filter, which upscales by a factor of two, to the blocks of pixel data.

14. The method according to claim 12 , wherein directionally interpolating the blocks of pixel data includes the sum of absolute difference metrics

M

45

=

∑

45

°

⁢

⁢

pairs

⁢

⁢

i

,

j

⁢

lum

i

-

lum

j

M

135

=

∑

135

°

⁢

⁢

pairs

⁢

⁢

i

,

j

⁢

lum

i

-

lum

j

of the pixel luminances.

15. The method according to claim 12 , wherein color sharpening the directional interpolated blocks of pixel data includes applying a non-linear sharpening filter to the directional interpolated blocks of pixel data.

16. The method according to claim 12 , wherein color sharpening the directional interpolated blocks of pixel data includes applying a low-pass filter.

17. The method according to claim 12 , wherein luminance clamping the color sharpened, directional interpolated blocks of pixel data includes clamping the color sharpened, directional interpolated blocks of pixel data to the luminance range of the received blocks of pixel data of the image.

18. The method according to claim 12 , further comprising:

directionally interpolating the block subpixel data, if the image scaling factor is equal to 2.0;

color sharpening the directional interpolated blocks of pixel data, if the image scaling factor is equal to 2.0;

luminance clamping the color sharpened, directional interpolated blocks of pixel data, if the image scaling factor is equal to 2.0; and

outputting the luminance clamped, color sharpened, directional interpolated blocks of pixel data as a scaled image, if the image scaling factor is equal to 2.0.

19. The method according to claim 12 further comprising:

directionally interpolating the blocks of pixel data, if the image scaling factor is greater than 2.0;

color sharpening the directional interpolated blocks of pixel data, if the image scaling factors greater than 2.0;

luminance clamping the color sharpened, directional interpolated blocks of pixel data, if the image scaling factor is greater than 2.0, to produce blocks of pixel data within inter-mediate image scaling;

iteratively performing the directional interpolating, color sharpening and luminance clamping, if the intermediate image scaling is not within the factor of 2 or less of the image scaling factor;

downsampling the directional interpolated blocks of pixel data with the intermediate image scaling, if the image scaling factor is greater than 1.0 and less than 2.0 of the image scaling factor,

color sharpening the downsampled, directional interpolated blocks of pixel data with the intermediate image scaling, if the scaling factor is greater than 1.0 or less than 2.0 of the image scaling factor;

luminance clamping the color sharpened, downsampled, directional interpolated blocks of pixel data with the intermediate image scaling, if the image scaling factor is greater than 1.0 and less than 2.0 of the image scaling factor; and

outputting the luminance clamped, color sharpened, down sampled, directional interpolated blocks of pixel data with the intermediate image scaling of a scaled image, if the image scaling factor is greater than 1.0 and less than 2.0 of the image scaling factor;

directionally interpolating the blocks pixel data with the intermediate image scaling, if the image scaling factor is equal to 2.0 of the image scaling factor;

color sharpening the directional interpolated blocks of pixel data with the intermediate image scaling, if the image scaling factor is equal to 2.0 of the image scaling factor;

luminance clamping the color sharpened, directional interpolated blocks of pixel data with the intermediate image scaling, if the image scaling factors equal to 2.0 of the image scaling factor; and

outputting the luminance clamped, color sharpened, directional interpolated blocks of pixel data with the intermediate image scaling as a scaled image, if the image scaling factors equal to 2.0 of the image scaling factor.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Aug 30, 2017
From: DONOVAN, WALTER
To: NVIDIA CORPORATION
Reel/Frame 043453/0006 →
Continuity (2)
Provisional Application 62044548 · Sep 2, 2014
Related Publication 20160063676A1 · Mar 3, 2016