Systems and methods for motion blur reduction

1. A method for compensating for perceived blur due to inter-frame motion in a sequence of digital video frames based on a direction and extent of the inter-frame motion, the method comprising:

generating a guess frame based on a current frame;

combining a plurality of blurring filters derived from a direction and extent of inter-frame motion to define a first blur-compensating filter, i.e. G N , wherein said first blur-compensation filter G N is independent of said guess frame;

convolving the first blur-compensating filter G N with the guess frame to directly produced a first output frame, i.e. O N , in the absence of any iterative updates to any intermediate image.

2. The method of claim 1 , wherein each of said blurring filters is defined by an inter-frame motion vector characterizing a direction and extent of inter-frame motion.

3. The method of claim 1 , wherein, said plurality of blurring filters are constructed from an initial blurring filter H, and said first blur-compensating filter G N simulates the effects of N iterative applications of said first blurring filter H.

4. The method of claim 3 , wherein said initial blurring filter H is a uniform one-dimensional blurring filter.

5. The method of claim 3 , wherein said first blur-compensating filter G N is a linear combination of said plurality of blurring filters.

6. The method of claim 3 , wherein first blur-compensating filter G N is defined as:

G

N

=

(

1

+

n

·

H

1

-

n

·

H

2

+

∑

i

=

3

2

⁢

n

⁢

(

α

n

-

1

,

i

-

α

n

-

1

,

i

-

2

)

·

H

i

)

for n=0 to N, α n,i denotes the coefficient associated with H i at a simulated iteration n, and α n,m =0 for i>2n.

7. The method of claim 3 , wherein pre-computed values of first blur-compensating filter G N for different values of N are stored in a memory, and the corresponding value of G N for a given value of N is accessed from said memory for convolving with guess frame in said step of convolving the first blur-compensating filter G N with the guess frame.

8. The method of claim 3 , wherein said first blur-compensating filter G N has an initial radius of r pixels, said method further including:

comparing said initial radius r with a predefined radius R, and conditionally truncating the radius r of first blur-compensating filter G N prior to said step of convolving the first blur-compensating filter G N with the guess frame, according to the following relation:

G

N

⁡

(

r

)

=

{

G

N

⁡

(

r

)

if

⁢

⁢



r



≤

R

0

otherwise

.

9. The method of claim 8 , wherein after truncating the radius r of first blur-compensating filter G N and prior to said step of convolving the first blur-compensating filter G N with the guess frame, adjusting the weights of the coefficients of G N to ensure they sum to one.

10. The method of claim 9 , wherein said predefined radius R is chosen to ensure that a majority of coefficients in G n are retained.

11. The method of claim 9 , wherein the pixel radius of said initial blurring filter H is R H , and said predefined radius R is set to four times R H .

12. The method of claim 11 , wherein the coefficients of the truncated first blur-compensating filter G n are pre-computed and stored in a look-up table, and are accessed from the look-up table for use in said step of convolving the first blur-compensating filter G N with the guess frame.

13. The method of claim 8 , further including following the step of convolving the first blur-compensating filter G N with the guess frame;

comparing each convolved pixel with a respective pixel in the current frame to generate an intensity adjustment for each respective pixel;

adjusting each pixel in the guess frame with its respective intensity adjustment to generate a blur-compensated frame; and

redefining the first output frame O N as said a blur-compensated frame.

14. The method of claim 1 , further including:

convolving the guess frame with a second first blur-compensating filter to produce a second output frame;

weighing said initial guess frame with a first weight, i.e. w1;

weighing said first output frame with a second weight, i.e. w2;

weighing said second output frame with a second weight, i.e. w3

combining the weighted guess frame, weighted first output frame, and weighted second output frame to create a composite frame;

convolving the composite frame with said first blur-compensating filter G N to produce a first blur-compensated frame.

15. The method of claim 14 , wherein said second blur-compensating filter is a uniform low-pass filter, said uniform low-pass filter being broader than said first blur-compensating filter.

16. The method of claim 14 , wherein the step to create said composite frame includes multiplying said composite frame by an adjustment step size β, and the values of adjustment step size β, first weight w1, second weight w2, and third weight w3 are selected to satisfy the following relationship: β·[w 1 −w 2 +w 3 ]=1.

17. The method of claim 14 , further including:

adding a weighted guess frame to said first blur-compensated frame to produce a second blur-compensated frame.

18. The method of claim 17 , wherein said weighted guess frame is defined weighing said guess frame with a fourth weight α, where α a is a tunable factor between zero and one, and satisfies the following relationship:

β·[ w 1 −w 2 +w 3 ]=(1−α).

19. The method as recited in claim 14 , wherein at least one of said first weight, second weight and third weight is calculated as a function of an estimated background luminance of the guess frame.