System and method for processing video images

• US 20060061583A1
• Filed: 09/23/2004
• Published: 03/23/2006
• Est. Priority Date: 09/23/2004
• Status: Active Grant

• Current Assignee: Intellectual Discovery Co., Ltd.
• Original Assignee: Conversion Works Incorporated
• Inventors: Simpson, Todd, Spooner, David A.
• Granted Patent: US 7,542,034 B2
• Time in Patent Office: Days
• US Class Current: 345/582
• CPC Class Codes: G06T 15/20 Perspective computation
• CPC Class Codes: H04N 13/261 with monoscopic-to-stereoscopic image conversion
• CPC Class Codes: H04N 13/286 having separate monoscopic and stereoscopic modes

Abstract

Some representative embodiments are directed to creating a “virtual world” by processing a series of two dimensional images to generate a representation of the physical world depicted in the series of images. The virtual world representation includes models of objects that specify the locations of the objects within the virtual world, the geometries of the objects, the dimensions of the objects, the surface representation of the objects, and/or other relevant information. By developing the virtual world representation, a number of image processing effects may be applied such as generation of stereoscopic images, object insertion, object removal, object translation, and/or other object manipulation operations.

Claims

1. A method for processing a video sequence, comprising:

defining graphical data associated with an object in multiple frames of said video sequence;

associating a three dimensional model with said object;

defining transforms experienced by said object between said multiple frames;

generating texture map data for a first frame of said multiple frames, wherein said generating (i) identifies a first portion of said three dimensional model visible according to a camera position associated with said first frame, (ii) includes defined graphical data associated with said first frame within texture map data for said first portion, (iii) identifies a second portion of said three dimensional model that is not visible according to a camera position associated with said first frame, (iv) processes defined graphical data associated with at least a second frame of said multiple frames according to the defined transforms experienced by said object between said first frame and said at least second frame, and (v) includes said processed graphical data within texture map data for said second portion.

2. The method of claim 1 wherein said defining transforms comprises:

defining changes in position experienced by said object.

3. The method of claim 1 wherein said defined transforms include at least one transform selected from the list consisting of:

a scaling transform, a rotational transform, and a morphing transform.

4. The method of claim 1 wherein said defining graphical data comprises:

defining outlines of said object in said multiple frames.

5. The method of claim 4 wherein said defining outlines comprises:

receiving input from a user via a user interface to define a plurality of points on an exterior of said object; and

executing an image processing algorithm to define edges between said plurality of points.

6. The method of claim 1 further comprising:

generating stereoscopic images for said first frame, wherein said generating stereoscopic images comprises rendering said three dimensional model using said generated texture map data from two perspectives.

7. The method of claim 1 further comprising:

repeating said generating texture map data for each of said multiple frames.

8. The method of claim 7 further comprising:

generating a sequence of stereoscopic images using said object model and said texture map data.

9. The method of claim 1 further comprising:

repeating said defining graphical data, associating, defining transforms, and generating for multiple objects in said multiple frames.

10. The method of claim 1 further comprising:

rendering a revised version of said first frame by removing another object that at least partially occludes said object associated with said three dimensional model in said first frame.

11. The method of claim 1 further comprising:

executing a camera reconstruction algorithm, wherein said generating processes defines graphical data associated with said second frame using data generated by said camera reconstruction algorithm.

12. The method of claim 1 wherein said generating prioritizes data for said second portion as a function of temporal distance from said first frame.

13. The method of claim 1 wherein said generating processes said identified graphical data using weighted averaging to compensate for image differences between said first and second frames.

14. A system for processing a video sequence, comprising:

a plurality of object models representing objects in said video sequence using three dimensional elements;

a plurality of transform data structures defining transforms experienced by said objects between said multiple frames;

a texture map generation algorithm for generating texture map data, wherein said algorithm, for each of said object models and for each of said multiple frames, is operable to (i) identify a first portion of a respective object model that is visible according to a camera position of a current frame of said multiple frames;

(ii) identify a second portion of said respective object model that is not visible according to said camera position;

(iii) include graphical data from said current frame in texture map data for said first portion for said current frame;

(iv) process graphical data from at least one frame other than said current frame according to a transform data structure; and

(v) include said processed graphical data in texture map data for said second portion.

15. The system of claim 14 wherein said plurality of object models encode data defining positions of said plurality of object models in three dimensions for said multiple frames.

16. The system of claim 14 wherein said plurality of object models encode data defining dimensions of said plurality of object models.

17. The system of claim 14 further comprising:

at least one camera data structure defining a characteristic of a camera used to capture said video sequence.

18. The system of claim 14 further comprising:

a rendering module for rending images using said plurality of object models and said texture map data.

19. The system of claim 17 further comprising:

a scene editing module for modifying object characteristics.

20. The system of claim 19 wherein said scene editing module receives user input to remove an object from a frame to be rendered.

21. The system of claim 19 wherein said scene editing module receives user input to insert an object into a frame to be rendered.

22. The system of claim 14 further comprising:

a rendering module for rendering images using said plurality of object models, said camera data, and said texture map data.

23. The system of claim 14 further comprising a module for reconstructing characteristics of a camera used to capture said video sequence.

24. The system of claim 14 further comprising a module for constructing a geometry representing objects in said video sequence.