Image processing method and apparatus for generating a 3D model of a target object

1. An image processing apparatus comprising:

a silhouette image generator to generate a three-dimensional (3D) silhouette image of a target object using both an input depth image and an input color image matched with the input depth image;

skeleton model generating unit to generate a skeleton model of the target object using the 3D silhouette image;

a 3D model generator to generate a 3D model of the target object by matching a template mesh model with the skeleton model of the target object using at least one processor; and

a 3D model refinement unit to refine the 3D model of the target object by refining the template mesh model matched with the skeleton model of the target object using the 3D silhouette image.

2. The image processing apparatus of claim 1 , wherein the target object corresponds to a human body.

3. The image processing apparatus of claim 1 , wherein the silhouette image generator divides the input depth image into a plurality of segments by performing a pixel segmentation of the input depth image, and generates the 3D silhouette image of the target image by performing a training-based head detection with respect to the plurality of segments.

4. The image processing apparatus of claim 3 , wherein the silhouette image generator refines the 3D silhouette image of the target object, generated using the input depth image, based on a result of performing the pixel segmentation of the input color image.

5. The image processing apparatus of claim 1 , wherein the motion modeling unit comprises:

a first calculator to perform a training-based head detection in the 3D silhouette image, and to calculate locations of limbs of the target object and a location of at least one joint that connects the plurality of limbs using a rectangle fitting algorithm based on a detected head portion; and

a second calculator to perform motion capturing of the target object based on the locations of the limbs and the location of the at least one joint, and to generate the skeleton model of the target object.

6. The image processing apparatus of claim 1 , further comprising:

a texture generator to generate a texture of the 3D model of the target object using the input color image.

7. The image processing apparatus of claim 6 , further comprising:

a texture database to store texture data applicable to the 3D model of the target object,

wherein the texture generator generates the texture of the 3D model of the target object based on the texture data stored in the texture database.

8. The image processing apparatus of claim 1 , further comprising:

an image matching unit to match the input color image with the input depth image when the input depth image and the input color image do not match each other.

9. The image processing apparatus of claim 1 , wherein the 3D model generator transforms the template mesh model to the skeleton model of the target object using a radial basis function.

10. An image processing apparatus comprising:

a silhouette image generator to generate a 3D silhouette image of a target object using both an input depth image and an input color image matched with the input depth image;

a skeleton model generating unit to generate a skeleton model of the target object using the 3D silhouette image;

a 3D model generator to generate a 3D model of the target object by matching a template mesh model with the skeleton model of the target object;

a 3D model refinement unit to refine the 3D model of the target object by refining the template mesh model matched with the skeleton model of the target object using the 3D silhouette image;

a texture database to store texture data applicable to the 3D model of the target object; and

a texture generator to generate a texture of the 3D model of the target object based on the texture data stored in the texture database using at least one processor.

11. The image processing apparatus of claim 10 , wherein the silhouette image generator divides the input depth image into a plurality of segments by performing a pixel segmentation of the input depth image, and generates the 3D silhouette image of the target image by performing a training-based head detection with respect to the plurality of segments.

12. The image processing apparatus of claim 10 , wherein the motion modeling unit comprises:

a first calculator to perform a training-based head detection with respect to the 3D silhouette image, and to calculate locations of limbs of the target object and a location of at least one joint that connects the plurality of limbs using a rectangle fitting algorithm based on a detected head portion; and

a second calculator to perform motion capturing of the target object based on the locations of the limbs of the target object and the location of the at least one joint, and to generate the skeleton model of the target object.

13. An image processing method comprising:

generating a 3D silhouette image of a target object using both an input depth image and an input color image matched with the input depth image;

generating a skeleton model of the target object using the 3D silhouette image;

generating a 3D model of the target object by matching a template mesh model with the skeleton model of the target object; and

refining the 3D model of the target object by refining the template mesh model matched with the skeleton model of the target object using the 3D silhouette image,

wherein the method is performed using at least one processor.

14. The image processing method of claim 13 , wherein the generating of the 3D silhouette image comprises:

dividing the input depth image into a plurality of segments by performing a pixel segmentation of the input depth image; and

generating the 3D silhouette image of the target image by performing a training-based head detection with respect to the plurality of segmentations.

15. The image processing method of claim 13 , wherein the generating of the skeleton model comprises:

performing a training-based head detection in the 3D silhouette image to calculate locations of limbs of the target object and a location of at least one joint that connects the plurality of limbs using a rectangle fitting algorithm based on a detected head portion; and

performing motion capturing of the target object based on the locations of the limbs of the target object and the location of the at least one joint to generate the skeleton model of the target object.

16. The image processing method of claim 13 , further comprising:

generating a texture of the 3D model of the target object using the input color image.

17. The image processing method of claim 13 , further comprising:

generating a texture of the 3D model of the target object based on texture data stored in a texture database.

18. At least one non-transitory computer readable medium storing computer readable instructions that control at least one processor to implement an image processing method comprising:

generating a 3D silhouette image of a target object using both an input depth image and an input color image matched with the input depth image

generating a skeleton model of the target object using the 3D silhouette image;

generating a 3D model of the target object by matching a template mesh model with the skeleton model of the target object; and

refining the 3D model of the target object by refining the template mesh model matched with the skeleton model of the target object using the 3D silhouette image.

19. An image processing method comprising:

generating a 3D silhouette image of a target object using both an input depth image and an input color image matched with the input depth image;

generating a skeleton model of the target object using the 3D silhouette image;

generating a 3D model of the target object by matching a template mesh model with the skeleton model of the target object;

refining the 3D model of the target object by refining the template mesh model matched with the skeleton model of the target object using the 3D silhouette image;

storing texture data applicable to the 3D model of the target object in a texture database; and

generating a texture of the 3D model of the target object based on the texture data stored in the texture database,

wherein the method is performed using at least one processor.

20. The image processing method of claim 19 , wherein the generating of the 3D silhouette image of the target object includes dividing the input depth image into a plurality of segments by performing a pixel segmentation of the input depth image, and generating the 3D silhouette image of the target image by performing a training-based head detection with respect to the plurality of segments.

21. The image processing method of claim 19 , wherein generating a 3D model of the target object comprises:

performing a training-based head detection with respect to the 3D silhouette image;

calculating locations of limbs of the target object and a location of at least one joint that connects the plurality of limbs using a rectangle fitting algorithm based on a detected head portion;

performing motion capturing of the target object based on the locations of the limbs of the target object and the location of the at least one joint; and

generating the skeleton model of the target object.

22. At least one non-transitory computer readable medium storing computer readable instructions that control at least one processor to implement the method of claim 19 .

23. The image processing apparatus of claim 1 , wherein the 3D model refinement unit adjusts the template mesh model according to the 3D silhouette image to align an outline of the template mesh model to correspond with an outline of the 3D silhouette image.

24. The image processing apparatus of claim 1 , wherein a posture of the template mesh model is different from a posture of the target object.