Methods and systems for reproducing a high resolution image from sample data

• US 6,023,535 A
• Filed: 08/30/1996
• Issued: 02/08/2000
• Est. Priority Date: 08/31/1995
• Status: Expired due to Term

• Current Assignee: Ricoh Company Limited
• Original Assignee: Ricoh Company Limited
• Inventors: Aoki, Shin
• Primary Examiner(s): Tran, Phuoc
• Assistant Examiner(s): Mariam, Daniel G.
• Application Number:
  US08/705,233
• Time in Patent Office: 1,257 Days
• Field of Search: 382/240, 382/268, 382/269, 382/299, 382/302, 382/304, 382/264
• US Class Current: 382/299
• CPC Class Codes: G06T 3/4053 based on super-resolution, i.e. the output image resolution being higher than the sensor resolution

Abstract

The current invention discloses methods of and systems for reproducing an original two dimensional image from data sampled at a low resolution. The data is sampled at the same frequency but at parallel displacement positions. The current invention discloses methods of and systems for substantially eliminating fold-over alias noises from the sampled data so that a high-resolution image is reproduced.

Claims

1. A method of reproducing a high-resolution original image from a plurality of low-resolution image data sets, comprising the steps of:

a) sampling an original image at a predetermined sampling resolution for a predetermined number of times so as to compile a predetermined number of the low-resolution image data sets, each of said sampled image data sets representing said original image at parallel displacement positions;

b) substantially eliminating fold-over aliasing noises caused by sampling the low-resolution image data sets during said step a), said step b) further comprising additional steps of;

h) for each sampled image data set, summing terms involving a product of each of said fold-over aliasing noises and a corresponding weight value;

i) adjusting said weight value so as to cancel said terms involving said fold-over aliasing noises with each other; and

c) reproducing the original image at least at an original resolution from the low-resolution image data sets after said step b).

2. The method of reproducing a high-resolution original image according to claim 1 wherein said original image includes pixels as units of an image, said parallel displacement positions being located within a distance of one pixel.

3. The method of reproducing a high-resolution original image according to claim 1 wherein said parallel displacement positions include vertically shifted positions.

4. The method of reproducing a high-resolution original image according to claim 1 wherein said parallel displacement positions include horizontally shifted positions.

5. The method of reproducing a high-resolution original image according to claim 1 wherein said predetermined resolution in said step a) is lower than that of said original image.

6. The method of reproducing a high-resolution original image according to claim 1 wherein said parallel displacement positions are known in said step a).

7. The method of reproducing a high-resolution original image according to claim 1 wherein said step a) further comprises the steps of:

d) obtaining a change in velocity in terms of a speed vector (p,q) based upon the following relation;

##EQU30## e) establishing a relative distance between a predetermined initial position and each of said parallel displacement positions based upon said p and said q; and

f) determining the relative positions of said parallel displacement positions with respect to said initial position.

8. The method of reproducing a high-resolution original image according to claim 1 wherein said b) further comprises the steps of:

g) filtering said sampled image data sets with a low pass filter, said low pass filter passing a signal having a frequency range of said original image prior to said sampling;

j) leaving one of said terms representing said original image; and

k) reconstructing said original image based upon said terms.

9. The method of reproducing a high-resolution original image according to claim 8 wherein said predetermined number of said image data sets is 2N+1 where N is a number of said fold over alias noises.

10. The method of reproducing a high-resolution original image according to claim 8 wherein said step g) is performed by executing the following multiplication:

space="preserve" listing-type="equation">E.sub.n (u,v)=LPF(u,v)D.sub.n (u,v)

where D_n is defined to be Σ

Σ

exp(-2π

i(p_n k+q_n l))F(u-k,v-l) summing over n as well as k and l, n being a number of said sampled image data sets, k and l being a center of said fold-over aliasing noise, P_n and q_n being a corresponding one of said parallel displacement positions, F(u,v) being a Fourier transfer of a function f(x, y) representing said original image, LPF(u,v) is a function representing said low pass filter, LPF(u,v)=1 if u is equal to or more than a negative value of the predetermined sampled frequency and v is equal to or less than a positive value of the predetermined sampled frequency, otherwise LPF(u,v)=0.

11. The method of reproducing a high-resolution original image according to claim 10 wherein said step h) is accomplished by taking a sum of w_n E_n (u,v) over n where w_n is said corresponding weight value for said n-th sampled image data set.

12. The method of reproducing a high-resolution original image according to claim 11 wherein in said step i), said weight value is adjusted so that a first sum of w_n over n is 1 while simultaneously a second sum of w_n exp(2π

i(p_n k+q_n l)) is 0.

13. The method of reproducing a high-resolution original image according to claim 12 wherein said weight value is further adjusted so that a third sum of W_n² is minimal.

14. The method of reproducing a high-resolution original image according to claim 11 wherein in said step i), said weight value is adjusted so that a first sum of w_n over n is 1 while simultaneously a second sum of ∥

w_n exp(2π

i(p_n k+q_n l))∥

² and a third sum w_n² are minimal.

15. The method of reproducing a high-resolution original image according to claim 1 wherein said step a) further includes the following additional steps of:

l) selecting a high resolution factor;

m) generating additional image data sets based upon said resolution factor and said sampled image data sets; and

n) storing said additional image data sets along with said sampled image data sets, said stored additional image data sets and said stored sampled image data sets defining high resolution data sets.

16. The method of reproducing a high-resolution original image according to claim 15 wherein said b) further comprises the steps of:

o) filtering said high resolution data sets with a low pass filter, said low pass filter passing a signal having a frequency range of said original image prior to said sampling;

r) leaving one of said terms involving a function representing said original image; and

s) reconstructing said original image at said selected high resolution factor based upon said terms.

17. A system for reproducing a high-resolution original image from a plurality of low-resolution image data sets, comprising:

a sampling unit for sampling an original image at a predetermined sampling resolution for a predetermined number of times so as to compile a predetermined number of the low-resolution image data sets, each of said sampled image data sets representing said original image at parallel displacement positions;

a low pass filter connected to said sampling unit for filtering out a certain portion of said sampled data;

a processing unit connected to said low pass filter for substantially eliminating fold-over aliasing noises caused in said sampled image data sets by said sampling unit, said processing unit further comprising;

a weight determination unit for determining a weight value so as to cancel terms involving a product of each of said fold-over aliasing noises and said weight value; and

a summation unit connected to said weight determination unit for summing said terms involving said product of each of said fold-over aliasing noises and said weight value, said fold-over aliasing noises being mutually canceled by placing said weights on said fold-over aliasing noises according to said parallel displacement positions; and

a reproduction unit connected to said processing unit for reproducing the original image at least at an original resolution from the low-resolution image data sets.

18. The system for reproducing a high-resolution original image according to claim 17 wherein said sampling unit samples pixels of said original image, said parallel displacement positions being located within a distance of one pixel.

19. The system for reproducing a high-resolution original image according to claim 17 wherein said sampling unit samples said image data sets representing said original image at vertically displacement positions.

20. The system for reproducing a high-resolution original image according to claim 17 wherein said sampling unit samples said image data sets representing said original image at horizontally displacement positions.

21. The system for reproducing a high-resolution original image according to claim 17 wherein said sampling unit samples said image data sets at said predetermined resolution which is lower than that of said original image.

22. The system for reproducing a high-resolution original image according to claim 17 further comprises a displacement positions estimating unit for determining relative positions of said parallel displacement positions with respect to an initial position based upon a change in velocity in terms of a speed as vector (p,q) using a following relation:

##EQU31## said displacement position unit establishing a relative distance between the predetermined initial position and each of said parallel displacement positions based upon said p and said q.

23. The system for reproducing a high-resolution original image according to claim 17 wherein said low pass filter filters said sampled image data sets so as to select a signal having a frequency range of said original image.

24. The system for reproducing a high-resolution original image according to claim 17 wherein said low pass filter designated by LPF(u,v) outputs 1 if u is equal to or more than a negative value of the predetermined sampling resolution and v is equal to or less than a positive value of the predetermined sampling resolution, otherwise LPF(u,v)outputting 0.

25. The system for reproducing a high-resolution original image according to claim 24 wherein said processing unit executes the following multiplication:

space="preserve" listing-type="equation">E.sub.n (u,v)=LPF(u,v)D.sub.n (u,v)

where D_n is defined to be Σ

Σ

exp(-2π

i(p_n k+q_n l))F(u-k,v-l) summing over n as well as k and l, n being a number of said sampled image data sets, k and l being a center of said fold-over aliasing noise, P_n and q_n being a corresponding one of said parallel displacement positions, F(u,v) being a Fourier transfer of a function f(x, y) representing said original image.

26. The system for reproducing a high-resolution original image according to claim 25 wherein said processing unit determines an optimal weight value so as to cancel product terms involving said fold-over aliasing noises with each other, said processing unit sums terms involving a product of each of said fold-over aliasing noises and said weight value.

27. The system for reproducing a high-resolution original image according to claim 26 wherein said processing unit determines said optimal weight value so as to make a first sum over n of w_n 1 while simultaneously a second sum of w_n exp(2π

i(p_n k+q_n l)) is 0.

28. The system for reproducing a high-resolution original image according to claim 27 wherein said weight value is further adjusted so that a third sum of w_n² is minimal.

29. The system for reproducing a high-resolution original image according to claim 26 wherein said processing unit determines said weight value so that a first sum of w_n over n is 1 while simultaneously a second sum of ∥

exp(2π

i(p_n k+q_n l))∥

² and a third sum w_n² are minimal.

30. The system for reproducing a high-resolution original image according to claim 17 wherein said processing unit further comprises:

a resolution selector unit for selecting a high resolution factor;

a resolution increasing unit for generating additional image data sets based upon said resolution factor and said sampled image data sets; and

a buffer unit for storing said additional image data sets along with said sampled image data sets, said stored additional image data sets and said stored sampled image data sets defining high resolution data sets.

31. The system for reproducing a high-resolution original image according to claim 30 wherein said processing unit determines an optimal weight value so as to cancel terms involving said fold-over aliasing noises with each other, said processing unit sums said terms involving a product of each of said fold-over aliasing noises and a corresponding weight value for each of said high resolution data sets.

32. The system for reproducing a high-resolution original image according to claim 17 wherein said reproduction unit generates said original image on an image-carrying medium.

33. The system for reproducing a high-resolution original image according to claim 17 wherein said reproduction unit is a display monitor.