Method and apparatus for annotating multimedia data in a computer-aided manner

1. A method for computer-aided annotating of multimedia data including a sequence of digitized images, comprising:

analyzing, executed by a computer, the multimedia data to detect objects in the multimedia data;

assigning, executed by the computer, each object to a role, with role assigning being ascertained by processing context information representing at least one model of the multimedia data; and

assigning, executed by the computer, the objects to object classes, with each object class formed of classified objects identified as similar,

wherein a pair of the classified objects in one of the object classes are identified as similar using an object-specific similarity measure based on at least one of visual, phonetic and geometric characteristics of the classified objects.

2. The method as claimed in claim 1 , wherein the context information is ascertained from an information source different from the multimedia data.

3. The method as claimed in claim 1 , wherein an identity is ascertained as a further step for a respective role.

4. The method as claimed in claim 3 , further comprising ascertaining an identity for an ascertained role by processing the context information.

5. The method as claimed in claim 4 , further comprising assigning an ascertained identity to at least one of a corresponding object and a corresponding object class assigned to the ascertained role.

6. The method as claimed in claim 5 , wherein the context information includes information about identities involved in the multimedia data.

7. The method as claimed in claim 6 , wherein the context information includes statistical knowledge about a context of the multimedia data.

8. The method as claimed in claim 7 , further comprising ascertaining a frequency of occurrence of one of the objects in the multimedia data.

9. The method as claimed in claim 8 , wherein the context information includes logical limitations.

10. The method as claimed in claim 9 , wherein the context information includes data from databases containing information about the multimedia data.

11. A device for the computer-aided annotating of multimedia data comprising a sequence of digitized images, comprising:

means for analyzing, executed by a computer, the multimedia data to detect objects in the multimedia data;

means for assigning, executed by the computer, each object to a role, with role assigning being ascertained by processing context information representing at least one model of the multimedia data; and

means for assigning, executed by the computer, the objects to object classes, with each object class formed of classified objects identified as similar,

wherein a pair of the classified objects in one of the object classes are identified as similar using an object-specific similarity measure based on at least one of visual, phonetic and geometric characteristics of the classified objects.

12. The device as claimed in claim 11 , wherein the context information is ascertained from an information source different from the multimedia data.

13. The device as claimed in claim 11 , wherein an identity is ascertained as a further step for a respective role.

14. The device as claimed in claim 13 , further comprising ascertaining an identity for an ascertained role by processing the context information.

15. The device as claimed in claim 14 , further comprising assigning an ascertained identity to at least one of a corresponding object and a corresponding object class assigned to the ascertained role.

16. A device for the computer-aided annotating of multimedia data comprising a sequence of digitized images, comprising:

a memory at least temporarily storing the multimedia data; and

a processor programmed to analyze the multimedia data, detect objects in the multimedia data, assign each object to a role, with role assigning being ascertained by processing, executed by the processor, context information representing at least one model of the multimedia data, and assign the objects to object classes, with each object class formed of classified objects identified as similar,

wherein a pair of the classified objects in one of the object classes are identified as similar using an object-specific similarity measure based on at least one of visual, phonetic and geometric characteristics of the classified objects.