Ghosting artifact reduction in temporal noise filtering

1. A method of noise filtering of a digital video sequence to reduce ghosting artifacts, the method comprising:

computing motion values for pixels in a frame of the digital video sequence based on a reference frame;

computing blending factors for the pixels based on the motion values;

generating filtered output pixel values by applying the blending factors to corresponding pixel values in the reference frame and the frame, wherein selected filtered output pixel values are converged toward corresponding pixel values in the frame to reduce ghosting artifacts, wherein generating filtered output pixel values comprises incrementing or decrementing a filtered output pixel value by 1 to move the filtered output pixel value closer to a corresponding pixel value in the frame when the filtered output pixel value is equal to a corresponding pixel value in the reference frame; and

outputting the filtered frame.

2. The method of claim 1 , wherein the selected filtered output pixel values are instantly converged toward the corresponding pixel values in the frame.

3. The method of claim 1 , wherein the selected filtered output pixel values are step-wise converged toward the corresponding pixel values in the frame.

4. The method of claim 1 , wherein generating filtered output pixel values comprises:

replacing a filtered output pixel value with a corresponding pixel value in the frame when the filtered output pixel value is equal to a corresponding pixel value in the reference frame.

5. The method of claim 1 , wherein computing blending factors comprises:

computing a blending factor as a 0 +(1−a 0 )m(x,y,n)/T m when (2 (a 0 +(1−a 0 ) m(x,y,n)/T m )m(x,y,n)>=3);

computing a blending factor as 1 when m(x,y,n)=T m or m(x,y,n)=0; and

computing a blending factor as 3/(2*(x,y,n)) otherwise,

wherein n is a number of the frame, x and y are pixel coordinates, m(x,y,n) is a motion value, T m is a motion threshold, and a 0 is a strength parameter.

6. The method of claim 1 , wherein computing blending factors comprises:

computing a blending factor as a 0 +(1−a 0 )m(x,y,n)/T m when (2 (a 0 +(1−a 0 ) m(x,y,n)/T m ) m(x,y,n)>=3) and m(x,y,n)<T m ; and

computing a blending factor as 1 otherwise,

wherein n is a number of the frame, x and y are pixel coordinates, m(x,y,n) is a motion value, T m is a motion threshold, and a 0 is a strength parameter.

7. The method of claim 1 , further comprising:

applying a spatial filter to the frame prior to computing the motion values.

8. A noise filter for filtering a digital video sequence to reduce ghosting artifacts, the noise filter comprising:

a frame buffer component configured to store an output frame of the noise filter to be used as a reference frame;

a motion detection component configured to compute motion values for pixels in a frame of the digital video sequence based on the reference frame;

a motion-to-blending factor conversion component configured to compute blending factors for the pixels based on the motion values; and

a frame blending component configured to generate filtered output pixel values by applying the blending factors to corresponding pixel values in the reference frame and the frame, wherein selected filtered output pixel values are converged toward corresponding pixel values in the frame to reduce ghosting artifacts; wherein the frame blending component is configured to perform one selected from a group consisting of incrementing or decrementing a filtered output pixel value by 1 to move the filtered output pixel value closer to a corresponding pixel value in the frame when the filtered output pixel value is equal to a corresponding pixel value in the reference frame, and replacing a filtered output pixel value with a corresponding pixel value in the frame when the filtered output pixel value is equal to a corresponding pixel value in the reference frame.

9. The noise filter of claim 8 , wherein the selected filtered output pixel values are instantly converged toward the corresponding pixel values in the frame.

10. The noise filter of claim 8 , wherein the selected filtered output pixel values are step-wise converged toward the corresponding pixel values in the frame.

11. The noise filter of claim 8 , wherein the motion-to-blending factor conversion component is configured to:

compute a blending factor as a 0 +(1−a 0 )m(x,y,n)/T m when (2 (a 0 +(1−a 0 ) m(x,y,n)/T m ) m(x,y,n)>=3);

compute a blending factor as 1 when m(x,y,n)=T m or m(x,y,n)=0; and

compute a blending factor as 3/(2*(x,y,n)) otherwise,

wherein n is a number of the frame, x and y are pixel coordinates, m(x,y,n) is a motion value, T m is a motion threshold, and a 0 is a strength parameter.

12. The noise filter of claim 8 , wherein the motion-to-blending factor conversion component is configured to:

compute a blending factor as a 0 +(1−a 0 )m(x,y,n)/T m when (2 (a 0 +(1−a 0 ) m(x,y,n)/T m ) m(x,y,n)>=3) and m(x,y,n)<T m ; and

compute a blending factor as 1 otherwise,

wherein n is a number of the frame, x and y are pixel coordinates, m(x,y,n) is a motion value, T m is a motion threshold, and a 0 is a strength parameter.

13. The noise filter of claim 8 , further comprising:

a spatial filter component configured to apply a spatial filter to the frame.

14. A digital system configured to filter a digital video sequence to reduce ghosting artifacts, the digital system comprising:

means for storing a filtered output frame to be used as a reference frame; and

means for computing motion values for pixels in a frame of the digital video sequence based on the reference frame;

means for computing blending factors for the pixels based on the motion values; and

means for generating filtered output pixel values by applying the blending factors to corresponding pixel values in the reference frame and the frame, wherein selected filtered output pixel values are converged toward corresponding pixel values in the frame to reduce ghosting artifacts; wherein the means for generating filtered output pixel values performs one selected from a group consisting of incrementing or decrementinq a filtered output pixel value by 1 to move the filtered output pixel value closer to a corresponding pixel value in the frame when the filtered output pixel value is equal to a corresponding pixel value in the reference frame, and replacing a filtered output pixel value with a corresponding pixel value in the frame when the filtered output pixel value is equal to a corresponding pixel value in the reference frame.

15. The digital system of claim 14 , wherein the means for computing blending factors:

computes a blending factor as a 0 +(1−a 0 )m(x,y,n)/T m when (2 (a 0 +(1−a 0 ) m(x,y,n)/T m ) m(x,y,n)>=3);

computes a blending factor as 1 when m(x,y,n)=T m or m(x,y,n)=0; and

computes a blending factor as 3/(2*(x,y,n)) otherwise,

wherein n is a number of the frame, x and y are pixel coordinates, m(x,y,n) is a motion value, T m is a motion threshold, and a 0 is a strength parameter.

16. The digital system of claim 14 , wherein the means for computing blending factors:

computes a blending factor as a 0 +(1−a 0 )m(x,y,n)/T m when (2 (a 0 +(1−a 0 ) m(x,y,n)/T m ) m(x,y,n)>=3) and m(x,y,n)<T m ; and

computes a blending factor as 1 otherwise,

wherein n is a number of the frame, x and y are pixel coordinates, m(x,y,n) is a motion value, T m is a motion threshold, and a 0 is a strength parameter.

17. The digital system of claim 14 , further comprising:

means for applying a spatial filter to the frame.