System, method, and apparatus for smoothing of edges in images to remove irregularities

Patent metadata

Field	Value
Publication number	US-9305337-B2
Application number	US-201314054575-A
Country	US
Kind code	B2
Filing date	Oct 15, 2013
Priority date	Sep 4, 2008
Publication date	Apr 5, 2016
Grant date	Apr 5, 2016

Abstract

System, method, and apparatus for smoothing of edges in images to remove irregularities are disclosed. In one aspect of the present disclosure, a method of image processing includes, identifying an edge in an image having an associated set of edge characteristics, determining the associated set of edge characteristics, and applying a low pass filter to a pixel of the edge based on the associated set of edge characteristics to generate a second image based on the image, wherein the edge in the image is smoothed in the second image. The method further includes generating a third image which is a blend of the original image and the second (edge-smoothed) image based on the associated set of edge characteristics.

First claim

What is claimed is:

1. A method of image processing, comprising:
applying an initial low pass filter to a first image to reduce noise and high frequency variations prior to edge smoothing;
identifying an edge in the first image, the edge having an associated set of edge characteristics, the associated set of edge characteristics comprising a gradient magnitude of the edge and an edge direction of the edge;
determining the gradient magnitude;
determining the edge direction;
assigning a bin direction to the edge direction, wherein bin boundaries are center lines between lines that connect pixel to pixel;
applying a low pass filter directionally to a pixel of the edge in the assigned bin direction to generate a second image based on the image, wherein the edge in the first image is smoothed in the second image; and
generating a third image based on the second image and the first image, wherein the third image is generated from blending a portion of the second image having the smoothed edge and a portion of the first image.

2. The method of claim 1 , further comprising processing binned edge direction values to remove or reduce number of isolated instances of edge angles.

3. The method of claim 1 , further comprising processing binned edge direction values such that if the pixel to the left and the pixel to the right of a certain pixel have the same edge direction value, then the edge direction value of the pixels to the left and to the right are assigned to the certain pixel.

4. The method of claim 1 , further comprising processing binned edge direction values such that if the pixel above and the pixel below a certain pixel have the same edge direction value, then the edge direction value of the pixels above and below are assigned to the certain pixel.

5. The method of claim 1 , wherein determining the edge direction comprises applying a low-pass filter to the horizontal gradient magnitudes.

6. The method of claim 1 , wherein determining the edge direction comprises applying a low-pass filter to the vertical gradient magnitudes.

7. The method of claim 1 , wherein determining the edge direction comprises applying a median filter to the sign of the gradient vector.

8. The method of claim 1 , further comprising assigning a confidence level to each of a plurality of pixels located on the edge, the confidence level corresponding to accuracy of the edge direction that is detected.

9. The method of claim 8 , wherein a multiplier for the blending of the second image and the first image is a product of a pixel confidence level regarding the edge direction and a comparison of a gradient magnitude of the edge of a threshold value.

10. A system, comprising:
a memory controller coupled to a memory, wherein the memory controller controls access to the memory; and
an image processor coupled to the memory, wherein the image processor executes an algorithm that:
identifies an edge in an image;
determines the gradient magnitude of the edge;
determines the edge direction of the edge;
assigns a bin direction to the edge direction determined for the edge;
applies a low pass filter directionally to a pixel of the edge based on the edge direction to generate a second image based on the image, wherein the edge in the first image is smoothed in the second image; and
generates a third image based on the second image and the first image, wherein the third image is generated from blending a portion of the second image having the smoothed edge and a portion of the first image.

11. The system of claim 10 , wherein the algorithm further processes binned edge direction values to remove or reduce number of isolated instances of edge angles.

12. The system of claim 10 , wherein the algorithm further processes binned edge direction values such that if the pixel to the left and the pixel to the right of a certain pixel have the same edge direction value, then the edge direction value of the pixels to the left and to the right are assigned to the certain pixel.

13. The system of claim 10 , wherein the algorithm further processes binned edge direction values such that if the pixel above and the pixel below a certain pixel have the same edge direction value, then the edge direction value of the pixels above and below are assigned to the certain pixel.

14. The system of claim 10 , wherein determining the edge direction comprises applying a low-pass filter to the horizontal gradient magnitudes.

15. The system of claim 10 , wherein determining the edge direction comprises applying a low-pass filter to the vertical gradient magnitudes.

16. The system of claim 10 , wherein determining the edge direction comprises applying a median filter to the sign of the gradient vector.

17. The system of claim 10 , wherein the algorithm further assigns a confidence level to each of a plurality of pixels located on the edge, the confidence level corresponding to accuracy of the edge direction that is detected.

18. The system of claim 17 , wherein a multiplier for the blending of the second image and the first image is a product of a pixel confidence levels regarding the edge direction and a comparison of a gradient magnitude of the edge to a threshold value.

Assignees

• Lattice Semiconductor Corp

Inventors

• Adams Dale R

Classifications

• G06T5/50

  using two or more images, e.g. averaging or subtraction · CPC title

• G06T2207/20192

  Edge enhancement; Edge preservation · CPC title

• G06T7/0085

  Physics · mapped topic

• G06T2207/20221

  Image fusion; Image merging · CPC title

• G06T3/403

  Edge-driven scaling; Edge-based scaling · CPC title