Structure and method for film grain noise reduction

• US 6,067,125 A
• Filed: 05/15/1997
• Issued: 05/23/2000
• Est. Priority Date: 05/15/1997
• Status: Expired due to Fees

• Current Assignee: Minerya System Incorporated
• Original Assignee: Minerya System Incorporated
• Inventors: May, William D.
• Primary Examiner(s): Powell, Mark R.
• Assistant Examiner(s): Sajous, Wesner
• Application Number:
  US08/857,084
• Time in Patent Office: 1,104 Days
• Field of Search: 348/606, 348/607, 348/608, 348/610, 348/618, 348/619, 348/620, 348/97, 348/459, 348/621, 348/451, 348/452, 348/415, 348/416
• US Class Current: 348/607
• CPC Class Codes: H04N 19/80 Details of filtering operations specially adapted for video compression, e.g. for pixel interpolation H04N19/635, H04N19/86 take precedence
• CPC Class Codes: H04N 5/21 Circuitry for suppressing or minimising disturbance, e.g. moiré or halo

Abstract

A video signal is both spatially and temporally filtered. The output video signal is temporally noise reduced in portions of the image which exhibits little or no motion, and is spatially noise reduced in portions where motion exists. This approach is particularly applicable to reducing film grain noise in video material that originates from images on photographic films, since film grain noise is more correlated spatially than temporally.

Claims

1. A system for processing a video signal, comprising:

a temporal filter receiving said video signal and providing a temporal noise reduced video signal and a motion signal indicative of motion in said video signal;

a spatial filter receiving said video signal and providing a spatial noise reduced video signal; and

a weighted average circuit receiving said motion signal, said temporal noise reduced video signal and said spatial noise reduced video signal, said weighted average circuit combining said temporal noise reduced video signal and said spatial noise reduced video signal according to said motion signal to provide an output video signal, said output video signal being weighted towards said temporal noise reduced video signal, when said motion signal indicates substantially no motion, and being weighted towards said spatial noise reduced video signal, otherwise.

2. A system as in claim 1, wherein said spatial filter utilizes local means and variances.

3. A system as in claim 1, wherein said spatial filter comprises an adaptive Wiener filter.

4. A system as in claim 3, wherein said adaptive Wiener filter includes a piecewise linear approximation.

5. A system as in claim 1, wherein said spatial filter takes into account pixels of said video signal configured in a 5-pixel kernel.

6. A system as in claim 1, wherein said temporal filter computes a weighted average of pixels in two successive frames.

7. A system as in claim 1, wherein said motion signal represents a difference between pixels in successive frames, and wherein said temporal noise reduced video signal and said spatial noise reduced video signal are weighted in said output video signal non-linearly with respect to said difference.

8. A method for processing a video signal, comprising:

temporally filtering said video signal to provide a temporal noise reduced video signal and a motion signal indicative of motion in said video signal;

spatially filtering said video signal to provide a spatial noise reduced video signal; and

providing a weighted average of said temporal noise reduced video signal and said spatial noise reduced video signal according to said motion signal to provide an output video signal, said output video signal being weighted towards said temporal noise reduced video signal, when said motion signal indicates substantially no motion, and being weighted towards said spatial noise reduced video signal, otherwise.

9. A method as in claim 8, wherein said step of spatially filtering utilizes local means and variances.

10. A method as in claim 8, wherein said step of spatially filtering comprises the step of adaptive Wiener filtering.

11. A method as in claim 10, wherein said adaptive Wiener filtering step includes an approximation step using a piecewise linear approximation.

12. A method as in claim 8, wherein said spatially filtering step takes into account pixels of said video signal configured in a 5-pixel kernel.

13. A method as in claim 8, wherein said temporally filtering step computes a weighted average of pixels in two successive frames.

14. A method as in claim 8, wherein said motion signal represents a difference between pixels in successive frames, and wherein said temporal noise reduced video signal and said spatial noise reduced video signal are weighted in said output video signal non-linearly with respect to said difference.