Edge detection

1. A method performed by a processing unit, for computing pixels along object edges in producing a deinterlaced picture from an interlaced video source, the method comprising:

calculating a gradient of image intensity for a 2-dimensional image represented by a plurality of known pixel values,

identifying an edge within the image using the gradient calculations,

identifying an angle of the edge using the gradient calculations,

selecting a bin that encompasses the edge angle, the bin having a high boundary with at least one known pixel value and a low boundary with at least one known pixel value,

calculating a distance from one boundary of the bin to the edge angle, and

blending the at least one known pixel value along the high boundary with the at least one known pixel value along the low boundary to calculate at least one new pixel, the blending being based on the calculated distance from the one bin boundary to the edge angle.

2. The method of claim 1 , wherein the act of calculating a gradient further comprises:

performing a vertical interpolation on the 2-dimensional image to produce an image upon which the gradient is calculated.

3. The method of claim 2 , wherein the act of calculating a gradient further comprises:

calculating a horizontal gradient of image intensity for the 2-dimensional image, and

calculating a vertical gradient of image intensity for the 2-dimensional image.

4. The method of claim 3 , wherein the act of identifying the angle of the edge using the gradient calculations further comprises:

calculating an edge angle using an inverse-tangent of the horizontal gradient calculation and the vertical gradient calculation.

5. The method of claim 4 , wherein the act of identifying an edge within the image using the gradient calculations further comprises:

combining the horizontal gradient calculation and the vertical gradient calculation to calculate a gradient magnitude which is used to determine whether the detected edge is valid.

6. The method of claim 1 , further comprising:

dynamically setting a threshold value using a local dynamic range,

comparing a magnitude of the gradient to the threshold value,

if the magnitude of the gradient is greater than the threshold value, setting the edge location as valid.

7. The method of claim 1 , further comprising:

subjecting a sign of the gradient to a median filter in order to remove noise thereby substantially eliminating incorrect edge determinations.

8. The method of claim 5 , further comprising:

calculating the horizontal gradient, the vertical gradient, the edge angle, the gradient magnitude, the distance from one boundary of the bin to the edge angle and the at least one new pixel using the processing unit,

using a memory coupled to the central processing unit to temporarily store the calculations,

validating the at least one new pixel calculation using a confidence level indicator, and

displaying the 2-dimensional image with the at least one new pixel on a screen.

9. The method of claim 3 , further comprising:

counting a number of angles present in the immediate area of the pixel,

counting a number of angles that are similar to the detected edge,

counting a number of angles that are different that the detected edge,

counting a number of pixels with similar directions as the detected edge,

counting a number of pixels with different directions as the detected edge,

combining the counting results to determine a confidence level,

blending edge pixel calculation with a vertical interpolation calculation to generate a first output pixel calculation, the blending being based on the confidence level, and

blending the first output pixel calculation with weaving calculation to generate a second output pixel calculation.

10. The method of claim 1 , wherein the boundaries of the bin are determined by an angle location of known pixel data.

11. A system for computing pixels along object edges in producing a deinterlaced picture from an interlaced video source, the system, comprising:

a processor;

a memory coupled to the processor;

wherein, instantiated on the processor and memory are:

means for calculating a gradient of image intensity for a 2-dimensional image,

means for identifying an edge within the image using the gradient calculations,

means for identifying an angle of the edge using the gradient calculations,

means for selecting a bin that encompasses the edge angle, the bin having a high boundary with at least one known pixel value and a low boundary with at least one known pixel value,

means for calculating a distance from one boundary of the bin to the edge angle, and

means for blending the at least one known pixel value along the high boundary with the at least one known pixel value along the low boundary to calculate at least one new pixel, the blending being based on the calculated distance from the one bin boundary to the edge angle.

12. The system of claim 11 , further comprising:

means for performing a vertical interpolation to calculate the gradient.

13. The system of claim 11 , further comprising:

means for calculating a horizontal gradient of image intensity for a 2-dimensional image, and

means for calculating a vertical gradient of image intensity for a 2-dimensional image.

14. The system of claim 11 , further comprising:

means for identifying an edge using a horizontal gradient calculation and a vertical gradient calculation.

15. The system of claim 11 , further comprising:

means for identifying an edge angle using a horizontal gradient calculation and a vertical gradient calculation.

16. The system of claim 11 , further comprising:

means for dynamically setting a threshold value using a dynamic range,

means for comparing the threshold value to a gradient magnitude, and

means for setting the edge location as valid if the gradient magnitude is greater than the threshold value.

17. The system of claim 11 , further comprising:

a median filter for removing noise.

18. The system of claim 11 , further comprising:

means for validating the pixel calculation using a confidence level indicator;

means for blending edge pixel calculation with a vertical interpolation calculation to generate a first output pixel calculation, the blending being based on a confidence level, and

means for blending the first output pixel calculation with weaving calculation to generate a second output pixel calculation.