Granted Patent B2

US 8,416,344 · App. 12/058,280 · Granted Apr 9, 2013

Abstract

Claims (18)

1. An iterative interpolation method for determining an interpolated video information value in an image that includes a number of image regions arrayed in matrix fashion, at least one of which is an image region to be interpolated, and that is disposed temporally between a first image and a second image of an image sequence, the method comprising:

a) specifying, by an image processing unit, a starting motion vector;

b) determining, by an image processing unit, for the starting motion vector a figure of merit, which depends on video information values of at least one group of image regions, this group of image regions comprising at least one image region in the first image, at least one image region in the second image and at least one image region in the image to be interpolated, the position of these image regions in the images being specified by the starting motion vector, wherein the figure of merit is the absolute value of a high-pass filter value that is obtained by applying a high-pass filter to the video information values;

c) selecting, by an image processing unit, at least one motion vector distinct from the starting motion vector and determining a figure of merit for this at least one motion vector;

d) comparing, by an image processing unit, the figures of merit for the starting motion vector and the at least one motion vector distinct from the starting motion vector, selecting the starting motion vector or the at least one other motion vector in dependence on the comparison result and determining an interpolated video information value using image regions that lie in the first and second images at positions specified by the selected motion vector or specified by a motion vector obtained by filtering the selected motion vector; and

e) repeating steps c) and d) at least once, the motion vector selected in the preceding step d) being employed as the starting motion vector, and the figure of merit of this selected motion vector as the figure of merit of the starting motion vector, when step c) is next performed.

2. The method of claim 1 , wherein steps c) and d) are repeated until a specified end condition has been attained.

3. The method of claim 2 , wherein the end condition has been attained when a specified number of iterations has been reached.

4. The method of claim 2 , wherein the end condition has been attained when the figure of merit associated with the interpolation direction in step d) satisfies a specified condition.

5. The method of claim 1 , wherein a new motion vector distinct from the starting interpolation direction is selected every time step c) is performed.

6. The method of claim 1 , wherein the at least one of the groups of image regions used in order to determine a figure of merit comprises the image region to be interpolated.

7. The method of claim 6 , wherein, before a first performance of step b), there is associated with the image region to be interpolated a starting interpolation value, which is generated by filtering of video information values of a group of image regions disposed in the first and second images at positions specified by the starting motion vector.

8. The method of claim 1 , wherein the at least one group of image regions for determining the figure of merit exhibits at least one further image region to be interpolated and wherein, before a first performance of step b) for the image region to be interpolated, step a) is performed for the at least one further image region to be interpolated, and wherein, before every repetition of steps c) and d) for the image region to be interpolated, steps c) and d) are performed for the at least one further image region to be interpolated.

9. The method of claim 8 , wherein steps c) and d) are performed for the at least one further image region to be interpolated only if the end condition for this further image region to be interpolated has not been attained.

10. The method of claim 1 , wherein at least two groups of image regions are employed in order to determine a figure of merit.

11. The method of claim 1 , wherein steps a) to d) are performed in parallel for a plurality of adjacent image regions and wherein the filtered motion vector for a first image region is obtained in step d) by filtering of the motion vector of the first image region and of at least one image region lying adjacent the first image region.

12. The method of claim 1 , wherein the only image regions employed in order to generate an interpolated video information value are original image regions.

13. The method of claim 1 , wherein the image regions are pixels and wherein the video information values are luminance values or chrominance values.

Assignments (9)