Digitized image stabilization using energy analysis method

• US 7,961,966 B2
• Filed: 01/04/2005
• Issued: 06/14/2011
• Est. Priority Date: 01/04/2005
• Status: Active Grant

• Current Assignee: Etron Technology Inc
• Original Assignee: Etron Technology Inc
• Inventors: Lu, Wen-Ming, Wang, Gyh-Bin
• Primary Examiner(s): Bali; Vikkram
• Assistant Examiner(s): Entezari; Michelle
• Application Number:
  US11/028,744
• Publication Number:
  US 20060146139A1
• Time in Patent Office: 2,352 Days
• Field of Search: 382/254, 348/208.99
• US Class Current: 382/254
• CPC Class Codes: H04N 23/68 for stable pick-up of the scene, e.g. compensating for camera body vibrations

Abstract

A method and an apparatus are provided for image stabilization for the output of analog-to-digital converters (ADC) and for phase-locked loops (PLL). The digital coding at the output of ADCs and PLLs is filtered by this method and apparatus to eliminate the noise which has contaminated the coding. The noise sources are noise picked up by the cable, system board noise, ADC power and ground noise paths, and switching noise. The differences of energy level of sequential pixels in the ADC and PLL digital outputs used in image displays are used to decide if correction is required. The method of image noise filtering is compatible with programmable circuitry. This allows the method to be tuned for optimal image stabilization.

Claims

1. A method of digitized image stabilization using energy analysis noise correction for analog-to-digital converters (ADC) comprising the steps of:

determining if a given image pixels'"'"' digital coding is not between the digital coding of its 2 adjacent pixels, determining if the absolute value of differences between a given image pixels digital coding and its two adjacent pixel'"'"'s digital coding is less than a pre-determined threshold value, and using said image pixels'"'"' digital coding and said absolute value of differences to select which pixels result in a stable image wherein said image is stabilized by calculating an optimum luminance using averaging and differences of light energy coding-

wherein G(x1), G(x2), and G(x3) are digital codings of 3 consecutive pixels x1, x2, x3,

wherein energy values, E1 and E2 are defined as

E1=G(x1)−

G(x2)

E2−

G(x3)−

G(x2) and wherein

Ethreshold is a variable threshold chosen to optimize image stabilization, and wherein,

If E1>

0, E2>

0, E1<

Ethreshold, and E2<

Ethreshold, then

G(x2)=[G(x1)+G(x3)]/2.

2. The method of digitized image stabilization using energy analysis of claim 1 wherein said image pixel'"'"'s digital coding is determined to not be between said coding of its two adjacent pixels in a monotonically increasing mode if the difference between a digital coding of a left-most adjacent pixel and said given image pixel is positive, and if the difference between the digital coding of the right-most adjacent pixel and said given image pixel is positive.

3. The method of digitized image stabilization using energy analysis of claim 1 wherein said image pixel'"'"'s digital coding is determined to not be between said coding of its two adjacent pixels in a monotonically decreasing mode if the difference between a digital coding of the left-most adjacent pixel and said given image pixel is negative, and if the difference between the digital coding of the right-most adjacent pixel and said given image pixel is negative.

4. The method of digitized image stabilization using energy analysis of claim 1 wherein said difference between said given image pixel'"'"'s digital coding'"'"'s absolute value and its two adjacent pixel'"'"'s digital coding'"'"'s absolute value is determined for possible digital coding correction, if said difference between said digital coding of said left-most adjacent pixel and said given pixel is less than said pre-determined threshold value, and if said difference between said digital coding of said right-most adjacent pixel and said given pixel is less than said pre-determined threshold value.

5. The method of digitized image stabilization using energy analysis of claim 1 wherein if it has been determined that both a given image pixel'"'"'s digital coding is not between the digital coding of its 2 adjacent pixels and a difference between given image pixel'"'"'s digital coding'"'"'s absolute value and its two adjacent pixel'"'"'s digital coding is less than a pre-determined threshold value, then said digital coding of said given image pixel is changed to an average of said digital coding of said two adjacent pixels.

6. The method of digitized image stabilization using energy analysis of claim 1 wherein if it has been determined that a given image pixel'"'"'s digital coding is not between the digital coding of its 2 adjacent pixels, and a difference between given image pixel'"'"'s digital coding'"'"'s absolute value and its two adjacent pixel'"'"'s digital coding is less than a pre-determined threshold value, then the digital coding of said given pixel is not changed and is kept to the original digital coding of said given image pixel.

7. The method of digitized image stabilization using energy analysis of claim 1 wherein said pre-determined threshold value is set to two or three least significant bits, in order to filter out ADC noise.

8. A method of digitized image stabilization using energy analysis noise correction for phase-locked-loops (PLL) comprising the steps of:

selecting an odd number, n, consecutive pixel samples which include a given image pixel, (n−

1)/2 consecutive image pixels which are adjacent on the left to said given image pixel, and (n−

1)/2 consecutive image pixels which are adjacent on the right to said given image pixel,

computing the n−

1 differences between digital codings of said n consecutive pixel samples, adding said n−

1 differences between said digital codings of said n consecutive pixel samples, to produce a total energy,

choosing a programmable, threshold for the summation of said n−

1 differences between said digital codings of said n consecutive pixel samples,

comparing said total energy to said threshold,

deciding if said total energy is greater than said threshold, and

changing said digital coding of said given image pixel if said energy is greater than said threshold, resulting in a stable image.

9. The method of digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 8 wherein said samples are chosen to provide an even number of transitions.

10. The method of digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 8 wherein said ‘

n’

consecutive pixel samples are chosen to be five to balance hardware complexity and effectiveness in representing alternating black and white images.

11. The method of digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 8 wherein said (n−

1) differences between digital codings of said n consecutive pixel samples are computed by subtracting a digital coding of a second pixel from a digital coding of a first pixel and a digital coding of a third pixel from said digital coding of said second pixel, wherein generally a digital coding of an x+1 pixel is subtracted from a digital coding of an x pixel, where x=1, 2, 3, . . . n.

12. The method of digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 11 wherein said n−

1 differences between said digital codings are added to produce a total energy value.

13. The method of digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 8 wherein said programmable threshold is chosen to distinguish smooth low contrast images from high contrast images such as alternating black and white images.

14. The method of digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 12 wherein said total energy value is compared to said programmable threshold.

15. The method of digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 14 where if said total energy value is greater than said programmable threshold, a “

change pixel coding flag is set.

16. The method of digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 15 wherein if said change pixel coding flag is set, said digital coding of said given image pixel is changed to a white value if said digital coding of said given image pixel is greater than a white threshold.

17. The method of digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 16 wherein if said change pixel coding flag is set, said digital coding of said given image pixel is changed to a black value if said digital coding of said given image pixel is less than said white threshold.

18. An apparatus for digitized image stabilization using energy analysis noise correction for analog-to digital converters (ADC) comprising:

a means for determining if a given image pixels'"'"' digital coding is not between the digital coding of its 2 adjacent pixels, a means for determining if the absolute value of differences between a given image pixel'"'"'s digital coding and its two adjacent pixels digital coding is less than a pre-determined threshold value, and

a means for using said image pixels'"'"' digital coding and said absolute value of differences to select which pixels result in a stable image, wherein said image is stabilized by calculating an optimum luminance using averaging and differences of light energy coding-

wherein G(x1), G(x2), and G(x3) are digital codings of 3 consecutive pixels x1, x2, x3,

wherein energy values, E1 and E2 are defined as

E1=G(x1)−

(x2)

E2−

G(x3)−

G(x2) and wherein

Ethreshold is a variable threshold chosen to optimize image stabilization, and wherein,

If E1>

0, E2>

0, E1<

Ethreshold, and E2<

Ethreshold, then

G(x2)=[G(x1)+G(x3)]/2.

19. The apparatus for digitized image stabilization using energy analysis of claim 18 wherein said image pixel'"'"'s digital coding is determined to not be between said coding of its two adjacent pixels in a monotonically increasing mode if the difference between a digital coding of a left-most adjacent pixel and said given image pixel is positive, and if the difference between the digital coding of the right-most adjacent pixel and said given image pixel is positive.

20. The apparatus for digitized image stabilization using energy analysis of claim 18 wherein said image pixel'"'"'s digital coding is determined to not be between said coding of its two adjacent pixels in a monotonically decreasing mode if the difference between a digital coding of the left-most adjacent pixel and said given image pixel is negative, and if the difference between the digital coding of the right-most adjacent pixel and said given image pixel is negative.

21. The apparatus for digitized image stabilization using energy analysis of claim 18 wherein said difference between said given image pixel'"'"'s digital coding'"'"'s absolute value and its two adjacent pixel'"'"'s digital coding'"'"'s absolute value is determined for possible digital coding correction, if said difference between said digital coding of said left-most adjacent pixel and said given pixel is less than said pre-determined threshold value, and if said difference between said digital coding of said right-most adjacent pixel and said given pixel is less than said pre-determined threshold value.

22. The apparatus for digitized image stabilization using energy analysis of claim 18 wherein if it has been determined that both a given image pixel'"'"'s digital coding is not between the digital coding of its 2 adjacent pixels and a difference between given image pixel'"'"'s digital coding'"'"'s absolute value and its two adjacent pixel'"'"'s digital coding is less than a pre-determined threshold value, then said digital coding of said given image pixel is changed to an average of said digital coding of said two adjacent pixels.

23. The apparatus for digitized image stabilization using energy analysis of claim 18 wherein if it has been determined that a given image pixel'"'"'s digital coding is not between the digital coding of its 2 adjacent pixels, and a difference between given image pixel'"'"'s digital coding'"'"'s absolute value and its two adjacent pixel'"'"'s digital coding is less than a pre-determined threshold value, then the digital coding of said given pixel is not changed and is kept to the original digital coding of said given image pixel.

24. The apparatus for digitized image stabilization using energy analysis of claim 18 wherein said pre-determined threshold value is set to two or three least significant bits, in order to filter out ADC noise.

25. A apparatus for digitized image stabilization using energy analysis noise correction for phase-locked-loops (PLL) comprising:

means for selecting an odd number, n, consecutive pixel samples which include a given image pixel, (n−

1)1/2 consecutive image pixels which are adjacent on the left to said given image pixel, and (n−

1)1/2 consecutive image pixels which are adjacent on the right to said given image pixel,

means for computing the n−

1 differences between digital codings of said n consecutive pixel samples,

means for adding said n−

1 differences between said digital codings of said n consecutive pixel samples, to produce a total energy,

means for choosing a programmable, threshold for the summation of said n−

1 differences between said digital codings of said n consecutive pixel samples, means for comparing said total energy to said threshold,

means for deciding if said total energy is greater than said threshold, and

means for changing said digital coding of said given image pixel if said energy is greater than said threshold, resulting in a stable image.

26. The apparatus for digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 25 wherein said samples are chosen to provide an even number of transitions.

27. The apparatus for digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 25 wherein said ‘

n’

consecutive pixel samples are chosen to be five to balance hardware complexity and effectiveness in representing alternating black and white images.

28. The apparatus for digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 25 wherein said (n−

1) differences between digital codings of said n consecutive pixel samples are computed by subtracting a digital coding of a second pixel from a digital coding of a first pixel and a digital coding of a third pixel from said digital coding of said second pixel, wherein generally a digital coding of an x+1 pixel is subtracted from a digital coding of an x pixel, where x=1, 2, 3, . . . n.

29. The apparatus for digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 28 wherein said n−

1 differences between said digital codings are added to produce a total energy value.

30. The apparatus for digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 25 wherein said programmable threshold is chosen to distinguish smooth low contrast images from high contrast images such as alternating black and white images.

31. The apparatus for digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 29 wherein said total energy value is compared to said programmable threshold.

32. The apparatus for digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 31 where if said total energy value is greater than said programmable threshold, a “

change pixel coding flag is set.

33. The apparatus for digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 32 wherein if said change pixel coding flag is set, said digital coding of said given image pixel is changed to a white value if said digital coding of said given image pixel is greater than a white threshold.

34. The apparatus for digitized image stabilization using energy analysis noise correction for phase-locked loops (PLL) of claim 33 wherein if said change pixel coding flag is set, said digital coding of said given image pixel is changed to a black value if said digital coding of said given image pixel is less than said white threshold.