Methods and apparatus to identify images in print advertisements

1. A method to automatically detect placement of an advertisement in a periodical or book comprising:

obtaining printed-media information representing a first printed periodical or printed book in which the advertisement appears, the printed-media information comprising at least one of source information, a publication name, a publication title, a publication date, a publisher name, or a page number of the first printed periodical or printed book;

computing, using a processor, a first image feature vector from the advertisement;

comparing the first image feature vector to a second image feature vector;

when the first image feature vector substantially matches the second image feature vector, storing the printed-media information in a database record associated with the second image feature vector, and

compiling a list of publications in which advertisements substantially matching the second image feature vector were printed based on the database record.

2. A method as defined in claim 1 , further comprising:

adding a new database record associated with the advertisement when the first image feature vector does not substantially match the second image feature vector;

collecting image information associated with the advertisement from an operator; and

storing the collected image information and the printed-media information in the new database record.

3. A method as defined in claim 1 , wherein comparing the first image feature vector to the second image feature vector comprises:

computing a first dissimilarity metric between the first image feature vector and the second image feature vector; and

determining that the first image feature vector matches the second image feature vector when the first dissimilarity metric is less than a threshold.

4. A method as defined in claim 1 , further comprising:

creating a set of one or more possible matching advertisements; and

presenting the one or more possible matching advertisements to an operator.

5. A method as defined in claim 4 , further comprising:

computing a dissimilarity metric between the first image feature vector and the second image feature vector; and

adding a second advertisement associated with the second image feature vector to the set of possible matching advertisements when the dissimilarity metric is less than a threshold.

6. A method as defined in claim 1 , further comprising:

scanning a cover page of the first printed periodical or printed book containing the advertisement; and

storing the digitized cover page in the database record.

7. A method as defined in claim 1 , further comprising detecting when the first and second image feature vectors match.

8. A method as defined in claim 7 , wherein detecting when the first and second image feature vectors match comprises:

computing a third image feature vector for the advertisement; and

comparing the third image feature vector with a fourth image feature vector associated with the second image feature vector.

9. A method as defined in claim 8 , wherein the third image feature vector is larger than the first image feature vector.

10. A method as defined in claim 8 , wherein the third image feature vector comprises a resized version of the advertisement.

11. An image identification system comprising:

an image feature extraction engine to compute an image feature vector for an image;

a database engine to:

compare the image with one or more additional images stored in a database based on the image feature vector;

when the image matches a first of the one or more additional images, store printed-media information representing a first printed periodical or printed book in which the image was presented in a database record associated with the first of the one or more additional images, wherein the printed-media information comprises at least one of source information, a publication name, a publication title, a publication date, a publisher name, or a page number for the first printed periodical or printed book; and

compile a list of publications in which advertisements substantially matching the second image feature vector were printed based on the database record.

12. An image identification system as defined in claim 11 , wherein the database record contains a list of printed media in which the image was presented.

13. An image identification system as defined in claim 11 , further comprising a user interface to allow an operator to indicate that the image partially matches the first of the one or more additional image, and to prompt the operator for updated image information in response to the partial match indication.

14. An image identification system as defined in claim 11 , further comprising a scanner to digitize a second image, the image feature extraction engine to compute the image feature vector in overlapping time intervals.

15. An image identification system as defined in claim 11 , further comprising a scanner to digitize a cover page of the first printed periodical or printed book containing the image, wherein the database engine is to store the digitized cover page in the database record.

16. An image identification system as defined in claim 15 , wherein the printed-media information is extracted from the digitized cover page.

17. A method as defined in claim 2 , wherein the image information comprises at least one of an advertiser name, a company name, a product name, a product version, a product type, a product price, a sale price, a page size, an image size, or a page location.

18. A method as defined in claim 6 , wherein obtaining the printed-media information comprising extracting the printed-media information from the scanned cover page.

19. A method as defined in claim 6 , wherein obtaining the printed-media information comprising receiving the printed-media information from a person.