Granted Patent B2

US 6,983,081 · App. 10/226,120 · Granted Jan 3, 2006

Abstract

Claims (26)

1. A method for integration of a source object into a base image with an area overlapping the source object, the method comprising the steps of:

identifying similar border pixels, that is, pixels on two sides of a border between the source object and the base image having differences from each other smaller than a first threshold;

calculating characteristic values of the similar border pixels in the source object and the base image using an energy function;

creating a tonal map using the characteristic values;

dividing the source object and the overlapped area in the base image into regions by segmentation filtering;

identifying similar regions in the source object, each of which, in the characteristic values, has a difference smaller than a second threshold from the similar border pixels in the source object and has a difference smaller than a third threshold from one of the regions in the overlapped area having a difference smaller than a fourth threshold from the similar border pixels in the overlapped area; and

applying the tonal map to the similar regions in the source object.

2. The method as claimed in claim 1 further comprising the step of:

blending the similar border pixels in the source object and base image by blending-filtering.

3. The method as claimed in claim 1 , wherein the differences and average pixels are calculated using RGB values of the pixels.

4. The method as claimed in claim 1 , wherein the differences and average pixels are calculated using HSV values.

5. The method as claimed in claim 1 , wherein the differences and average pixels are calculated using LAB values.

6. The method as claimed in claim 1 , wherein the characteristic values are averages.

7. A method for integration of a source object into a base image with an area overlapping the source object, the method comprising the steps of:

identifying groups of similar border pixels on two sides of a border between the source object and the base image, wherein, in each of the groups, the similar border pixels have differences from each other below a first threshold;

calculating characteristic values of the similar border pixels in the source object and the base image using an energy function for each of the groups;

creating a tonal map using the characteristic values for each of the groups;

dividing the source object and the overlapped area in the base image into regions by segmentation filtering;

identifying similar regions in the source object, each of which, in the characteristic values, has a difference smaller than a second threshold from a corresponding group of the similar border pixels in the source object and has a difference smaller than a third threshold from one of the regions in the overlapped area having a difference smaller than a fourth threshold from the corresponding group of the similar border pixels in the overlapped area; and

applying to each of the similar regions in the source object one of the tonal maps created for the corresponding group of the similar border pixels.

8. The method as claimed in claim 7 further comprising the step of:

blending the similar border pixels in the source object and base image by blending-filtering.

9. The method as claimed in claim 7 , wherein the differences and average pixels are calculated using RGB values of the pixels.

10. The method as claimed in claim 7 , wherein the differences and average pixels are calculated using HSV values.

11. The method as claimed in claim 7 , wherein the differences and average pixels are calculated using LAB values.

12. The method as claimed in claim 7 , wherein the characteristic values are averages.

Assignments (14)