Autofocus Method

• US 20120200725A1
• Filed: 02/03/2011
• Published: 08/09/2012
• Est. Priority Date: 02/03/2011
• Status: Active Grant

• Current Assignee: FotoNation Limited (Adeia, Inc.)
• Original Assignee: Tessera Technologies Ireland Limited (Adeia, Inc.)
• Inventors: Albu, Felix, Poenaru, Vlad, Drimbarean, Alexandru
• Granted Patent: US 8,508,652 B2
• Time in Patent Office: Days
• US Class Current: 348/222.1
• CPC Class Codes: H04N 23/673 based on contrast or high frequency components of image signals, e.g. hill climbing method
• CPC Class Codes: H04N 23/676 Bracketing for image capture at varying focusing conditions

Abstract

An autofocus method includes acquiring multiple images each having a camera lens focused at a different focus distance. A sharpest image is determined among the multiple images. Horizontal, vertical and/or diagonal integral projection (IP) vectors are computed for each of the multiple images. One or more IP vectors of the sharpest image is/are convoluted with multiple filters of different lengths to generate one or more filtered IP vectors for the sharpest image. Differences are computed between the one or more filtered IP vectors of the sharpest image and one or more IP vectors of at least one of the other images of the multiple images. At least one blur width is estimated between the sharpest image and the at least one of the other images of the multiple images as a minimum value among the computed differences over a selected range. The steps are repeated one or more times to obtain a sequence of estimated blur width values. A focus position is adjusted based on the sequence of estimated blur width values.

Claims

1. A method of automatically focusing a camera on a scene, comprising:

(a) acquiring multiple images each having a lens of the camera focused at a different focus distance;

(b) determining a sharpest image among the multiple images;

(c) computing horizontal, vertical or diagonal integral projection vectors, or combinations thereof, for each of the multiple images;

(d) convoluting one or more IP vectors of the sharpest image with multiple filters of different lengths to generate one or more filtered IP vectors for the sharpest image;

(e) computing differences between the one or more filtered IP vectors of the sharpest image and one or more IP vectors of at least one of the other images of the multiple images;

(f) estimating at least one blur width between the sharpest image and the at least one of the other images of the multiple images as a minimum value among the computed differences over a selected range;

(g) repeating (b)-(f) one or more times to obtain a sequence of estimated blur width values; and

(h) adjusting a focus position based on the sequence of estimated blur width values.

2. The method of claim 1, wherein the determining said sharpest image comprises a gradient image-based process.

3. The method of claim 1, wherein the differences comprise absolute sums of differences.

4. The method of claim 1, wherein the differences comprise sums of squared differences.

5. The method of claim 1, wherein the multiple filters of different lengths comprise integral projection vectors of blur kernels of different widths.

6. The method of claim 1, wherein the blur kernels comprise Gaussian or circular averaging kernels, or combinations thereof

7. The method of claim 1, wherein the minimum value indicates a suitable filter length.

8. The method of claim 1, further comprising downscaling or cropping lengths of the integral projection vectors in order to reduce complexity.

9. The method of claim 1, further comprising applying a descent process to reduce a number of computations.

10. The method of claim 1, further comprising computing errors at different interval lengths in order to avoid one or more local minima.

11. The method of claim 1, wherein the adjusting of the focus position comprises a smaller or greater adjustment, respectively, for smaller or greater estimated blur width values.

12. The method of claim 1, further comprising:

estimating a focus kernel difference between the sharpest image and the at least one of the other images of the multiple images; and

computing an approximate focus level position based on the estimated focus kernel difference.

13. The method of claim 12, further comprising refining the focus level position by acquiring one or more images close to the approximate focus level position and applying a gradient method to determine a best match.

14. A image acquisition device that automatically focuses on a scene, comprising:

(a) a lens and image sensor for acquiring multiple images each having the lens focused at a different focus distance;

(b) a processor;

(c) a memory having code embedded therein for programming the processor to perform an autofocus method, including;

(i) determining a sharpest image among the multiple images;

(ii) computing horizontal, vertical or diagonal integral projection vectors, or combinations thereof, for each of the multiple images;

(iii) convoluting one or more IP vectors of the sharpest image with multiple filters of different lengths to generate one or more filtered IP vectors for the sharpest image;

(iv) computing differences between the one or more filtered IP vectors of the sharpest image and one or more IP vectors of at least one of the other images of the multiple images;

(v) estimating at least one blur width between the sharpest image and the at least one of the other images of the multiple images as a minimum value among the computed differences over a selected range;

(vi) repeating (i)-(v) one or more times to obtain a sequence of estimated blur width values; and

(vii) adjusting a focus position based on the sequence of estimated blur width values.

15. The device of claim 14, wherein the determining said sharpest image comprises a gradient image-based process.

16. The device of claim 14, wherein the differences comprise absolute sums of differences.

17. The device of claim 14, wherein the differences comprise sums of squared differences.

18. The device of claim 14, wherein the multiple filters of different lengths comprise integral projection vectors of blur kernels of different widths.

19. The device of claim 14, wherein the blur kernels comprise Gaussian or circular averaging kernels, or combinations thereof

20. The device of claim 14, wherein the minimum value indicates a suitable filter length.

21. The device of claim 14, wherein the method further comprises downscaling or cropping lengths of the integral projection vectors in order to reduce complexity.

22. The device of claim 14, wherein the method further comprises applying a descent process to reduce a number of computations.

23. The device of claim 14, wherein the method further comprises computing errors at different interval lengths in order to avoid one or more local minima.

24. The device of claim 14, wherein the adjusting of the focus position comprises a smaller or greater adjustment, respectively, for smaller or greater estimated blur width values.

25. The device of claim 14, wherein the method further comprises:

estimating a focus kernel difference between the sharpest image and the at least one of the other images of the multiple images; and

computing an approximate focus level position based on the estimated focus kernel difference.

26. The device of claim 25, wherein the method further comprises refining the focus level position by acquiring one or more images close to the approximate focus level position and applying a gradient method to determine a best match.

27. One or more non-transitory, processor-readable devices having code embedded therein for programming a processor to perform a method of automatically focusing a camera or other camera-enabled device on a scene, the method comprising:

(a) acquiring multiple images each having a lens of a camera or other camera-enabled device focused at a different focus distance;

(b) determining a sharpest image among the multiple images;

(c) computing horizontal, vertical or diagonal integral projection vectors, or combinations thereof, for each of the multiple images;

(d) convoluting one or more IP vectors of the sharpest image with multiple filters of different lengths to generate one or more filtered IP vectors for the sharpest image;

(e) computing differences between the one or more filtered IP vectors of the sharpest image and one or more IP vectors of at least one of the other images of the multiple images;

(f) estimating at least one blur width between the sharpest image and the at least one of the other images of the multiple images as a minimum value among the computed differences over a selected range;

(g) repeating (b)-(f) one or more times to obtain a sequence of estimated blur width values; and

(h) adjusting a focus position based on the sequence of estimated blur width values.

28. The one or more non-transitory, processor-readable media of claim 27, wherein the determining said sharpest image comprises a gradient image-based process.

29. The one or more non-transitory, processor-readable media of claim 27, wherein the differences comprise absolute sums of differences.

30. The one or more non-transitory, processor-readable media of claim 27, wherein the differences comprise sums of squared differences.

31. The one or more non-transitory, processor-readable media of claim 27, wherein the multiple filters of different lengths comprise integral projection vectors of blur kernels of different widths.

32. The one or more non-transitory, processor-readable media of claim 27, wherein the blur kernels comprise Gaussian or circular averaging kernels, or combinations thereof

33. The one or more non-transitory, processor-readable media of claim 27, wherein the minimum value indicates a suitable filter length.

34. The one or more non-transitory, processor-readable media of claim 27, wherein the method further comprises downscaling or cropping lengths of the integral projection vectors in order to reduce complexity.

35. The one or more non-transitory, processor-readable media of claim 27, wherein the method further comprises applying a descent process to reduce a number of computations.

36. The one or more non-transitory, processor-readable media of claim 27, wherein the method further comprises computing errors at different interval lengths in order to avoid one or more local minima.

37. The one or more non-transitory, processor-readable media of claim 27, wherein the adjusting of the focus position comprises a smaller or greater adjustment, respectively, for smaller or greater estimated blur width values.

38. The one or more non-transitory, processor-readable media of claim 27, wherein the method further comprises:

estimating a focus kernel difference between the sharpest image and the at least one of the other images of the multiple images; and

computing an approximate focus level position based on the estimated focus kernel difference.

39. The one or more non-transitory, processor-readable media of claim 38, wherein the method further comprises refining the focus level position by acquiring one or more images close to the approximate focus level position and applying a gradient method to determine a best match.