System and method for forensic analysis of media works

1. A method, comprising:

aligning at least one suspect image frame of a suspect work with at least one reference image frame of a reference work, wherein aligning comprises:

temporally aligning the at least one suspect image frame with the at least one reference image frame; and

spatially aligning the at least one suspect image frame with the at least one reference image frame;

identifying one or more reference forensic marks in the at least one reference image frame, wherein the one or more reference forensic marks comprise at least one non-embedded forensic mark; and

determining, based on the at least one non-embedded forensic mark of the one or more reference forensic marks in the at least one reference image frame, whether the reference work corresponds to the suspect work.

2. The method of claim 1 , wherein determining whether the reference work corresponds to the suspect work comprises matching the at least one non-embedded forensic mark of the at least one reference image frame with one or more suspect forensic marks in the at least one suspect image frame.

3. The method of claim 2 , wherein matching the at least one non-embedded forensic mark of the at least one reference image frame with the one or more suspect forensic marks in the at least one suspect image frame comprises:

identifying the one or more suspect forensic marks in the at least one suspect image frame; and

comparing the at least one non-embedded forensic mark of the at least one reference image frame with the one or more suspect forensic marks identified in the at least one suspect image frame.

4. The method of claim 3 , wherein comparing the at least one non-embedded forensic mark of with the one or more suspect forensic marks comprises scoring a similarity between the at least one non-embedded forensic mark of the at least one reference image frame and the one or more suspect forensic marks identified in the at least one suspect image frame.

5. The method of claim 1 , wherein determining whether the reference work corresponds to the source of the suspect work comprises searching the at least one suspect image frame for correspondence with the at least one non-embedded forensic mark of the at least one reference image frame.

6. The method of claim 5 , wherein searching the at least one suspect image frame comprises determining a location of the at least one non-embedded forensic mark of the at least one reference image frame and searching a corresponding location of the at least one suspect image frame for features corresponding to the at least one non-embedded forensic mark of the at least one reference image frame.

7. The method of claim 1 , wherein temporally aligning comprises:

extracting low-level image features from the at least one reference image frame and from the at least one suspect image frame; and

aligning the low-level image features of the at least one reference frame with the low-level image features of the at least one suspect image frame.

8. The method of claim 7 , wherein the low-level image features comprise one or more of a color histogram change, a Minkowski distance, or a histogram intersection, or a combination thereof.

9. The method of claim 1 , wherein temporally aligning further comprises scene change detection, normalizing a sampling rate, or phase correlation of low-level features, or a combination thereof.

10. The method of claim 1 , wherein the at least one non-embedded forensic mark comprises imagery of a speck of dirt, a scratch, a perspective distortion, an audience interaction, or a combination thereof.

11. The method of claim 1 , wherein determining whether the reference work is the source of the suspect work comprises scoring a similarity between the at least one reference image frame and the at least one suspect image frame based on the at least one non-embedded forensic mark of the at least one reference image frame.

12. The method of claim 1 , wherein spatially aligning comprises minimization of disjoint information, maximization of mutual information, or optimization of metrics, or a combination thereof.

13. A method, comprising:

aligning reference image frames of a reference work with suspect image frames of a suspect work, wherein aligning comprises:

spatially aligning the suspect image frames with the reference image frames; and

temporally aligning the suspect image frames with the reference image frames;

identifying one or more reference forensic marks in the reference image frames, wherein the one or more reference forensic marks comprise at least one non-embedded forensic mark; and

scoring a similarity between the reference work and the suspect work based on the at least one non-embedded forensic mark of the one or more reference forensic marks identified in the reference image frames.

14. The method of claim 13 , wherein scoring the similarity between the reference work and the suspect comprises searching the suspect image frames for correspondence with the at least one non-embedded forensic mark in the reference image frames.

15. The method of claim 13 , wherein scoring the similarity between the reference work and the suspect work comprises:

identifying one or more suspect forensic marks in the suspect image frames, wherein the one or more suspect forensic marks comprise at least one additional non-embedded forensic mark; and

comparing the at least one non-embedded forensic mark of the reference image frames with the at least one additional non-embedded forensic mark of the suspect image frames.

16. The method of claim 13 , wherein temporally aligning comprises:

extracting low-level image features from the reference image frames of the reference work and from the suspect image frames of the suspect work; and

aligning the low-level image features of the suspect image frames with the low-level image features of the reference image frames.

17. The method of claim 13 , wherein temporally aligning further comprises scene change detection, normalizing a sampling rate, phase correlation of low-level features, or a combination thereof.

18. A system, comprising:

a processor;

at least one memory configured to store a reference work comprising reference image frames and a suspect work comprising suspect image frames;

an alignment section that, when executed by the processor, performs temporal alignment and spatial registration between the reference image frames and the suspect image frames;

a mark detection section that, when executed by the processor, identifies at least one reference forensic mark in the reference image frames and at least one suspect forensic mark in the suspect image frames, wherein the at least one reference forensic mark comprises at least one reference non-embedded forensic mark, and the at least one suspect forensic mark comprises at least one suspect non-embedded forensic mark; and

a mark comparison section that, when executed by the processor, determines, based on the at least one suspect non-embedded forensic mark and the at least one reference non-embedded forensic mark, whether the reference work corresponds to the suspect work.

19. The system of claim 18 , wherein temporally aligning comprises:

extracting low-level image features from the reference image frames of the reference work and from the suspect image frames of the suspect work; and

aligning the low-level image features of the suspect image frames with the low-level image features of the reference image frames.

20. The system of claim 18 , wherein temporally aligning further comprises change detection, normalizing a sampling rate, phase correlation of low-level features, or a combination thereof.