Claims
- 1. A system for identifying features comprising:
a processor; a storage device accessible by said processor arranged to store a data set comprising digital representations of a plurality of data objects; an input device coupled to said processor configured to accept input from a user; and, software executable by said processor for:
selecting a data subset from said data set; selecting at least one segment of interest from said data subset; providing at least one tag for said at least one segment of interest; constructing a transformation function based upon said at least one segment of interest; deriving at least one of a feature and a signature from said transformation function; and, outputting at least one of said feature, and said signature.
- 2. The system for identifying features according to claim 1, wherein said segment of interest is selected by said user interacting with said software.
- 3. The system for identifying features according to claim 1, wherein said segment of interest is selected automatically by said software.
- 4. The system for identifying features according to claim 1, wherein said software is further executable to process said data subset prior to deriving said feature to accentuate at least one aspect of interest.
- 5. The system for identifying features according to claim 1, wherein said software is further executable to identify a characteristic from said at least one segment of interest, wherein said transformation function is constructed based further upon said characteristic.
- 6. The system for identifying features according to claim 5, wherein said characteristic comprises identifying two or more segments of interest from the group consisting of similar, distinct, dissimilar, included, excluded, different, identical, mutually exclusive, related, unrelated, and segments that should be ignored.
- 7. The system for identifying features according to claim 1, wherein said software is configured to repeatedly derive new features based upon differently selected segments of interest, the derived features collectively defining a feature set.
- 8. The system for identifying features according to claim 1, wherein said transformation function is based upon at least one of a weighted mask, a percentile function, and at least one computation derived from at least one subsection of said segment of interest.
- 9. The system for identifying features according to claim 8, wherein said feature is based upon at least one of a correlation and a covariance computation of said subsection.
- 10. The system for identifying features according to claim 8, wherein said feature is based upon a distance between at least two subsections of said segments of interest.
- 11. The system for identifying features according to claim 1, wherein said transformation function is derived by:
breaking said segment of interest into a collection of subsections; and, performing a function that maps at least one subsection to at least one vector.
- 12. The system for identifying features according to claim 11, wherein said at least one vector is transformed into a signature by clustering said at least one vector across said data set, wherein each data object is characterized in a frequency table indicating how many vectors for that subset are in each cluster.
- 13. The system for identifying features according to claim 1, wherein said transformation function is derived by:
deconstructing said segments of interest into subsections expressed as 5I=⋃l∈ΛSegl, where I is the segment and Segl is a subsection of said segment of interest; and, letting f: Seg→Rk map a subsection to a vector.
- 14. The system for identifying features according to claim 11, wherein said data set comprises a collection of images, and said transformation function expands said at least one segment into the pixel gray.
- 15. The system for identifying features according to claim 1, wherein said transformation function is derived by:
gathering values of primitives; summarizing a distribution of said primitives; and, applying the summarization across at least a portion of said data set.
- 16. A system for identifying features comprising:
at least one processor; a storage device accessible by said processor arranged to store a data set comprising a plurality of data objects; an input device coupled to said processor and configured to accept input from a user; and, software executable by said at least one processor for:
selecting a data subset from said data set, wherein said data subset is selected by one of a user, said software, and a combination of said software and a user; selecting at least one segment of interest from said data subset, wherein said segment of interest is selected by one of a user, said software, and a combination of said software and a user; providing at least one tag for said at least one segment of interest, wherein said at least one tag is selected by one of a user, said software, and a combination of said software and a user; constructing a transformation function based upon said at least one segment of interest; deriving at least one of a feature and a signature from said transformation function; and, outputting at least one of said feature and said signature.
- 17. The system for identifying features according to claim 16, wherein said software is further executable to process said data subset prior to deriving said feature to accentuate at least one aspect of interest.
- 18. The system for identifying features according to claim 16, wherein said software is further executable to identify a characteristic from said at least one segment of interest, wherein said transformation function is constructed further based further upon said characteristic.
- 19. The system for identifying features according to claim 16, wherein said characteristic comprises identifying two or more segments of interest from the group consisting of similar, distinct, dissimilar, included, excluded, different, identical, mutually exclusive, related, unrelated, and segments that should be ignored.
- 20. The system for identifying features according to claim 16, wherein said software is further executable to perform operation comprising:
selecting additional regions of interest, deriving at least one of additional features and additional signatures therefrom; and, outputting at least one of said additional features and additional signatures.
- 21. A system for identifying features from a data set comprising:
means for selecting a data subset from said data set; means for selecting segments of interest from said data subset; means for providing tags for said segments of interest; means for assigning characteristics to said segments of interest; means for constructing a transformation functions based upon said segments of interest, said tags, and said characteristics; means for deriving features from said transformation functions; means for outputting said feature.
- 22. A system for identifying features from a data set comprising:
a storage device having a plurality of digital representations of data objects stored thereon; an input device configured to accept input from a user; a processor coupled to said storage device and said input device programmed to:
select a data subset from said plurality of digital representations of data objects; select at least one segment of interest from said data subset; provide at least one tag for said at least one segment of interest; construct a transformation function based upon said at least one segment of interest; derive at least one of a feature and a signature from said transformation function; and, output at least one of said feature and said signature.
- 23. The system for identifying features from a data set according to claim 22, wherein said segment of interest is selected by said user.
- 24. The system for identifying features from a data set according to claim 22, wherein said segment of interest is selected automatically by said processor.
- 25. The system for identifying features from a data set according to claim 22, wherein said processor is further operative to process said data subset prior to deriving said feature to accentuate at least one aspect of interest.
- 26. The system for identifying features from a data set according to claim 22, wherein said processor is further operative to identify a characteristic from said at least one segment of interest, wherein said transformation function is constructed based further upon said characteristic.
- 27. The system for identifying features from a data set according to claim 22, wherein said characteristic comprises identifying two or more segments of interest from the group consisting of similar, distinct, dissimilar, included, excluded, different, identical, mutually exclusive, related, unrelated, and segments that should be ignored.
- 28. The system for identifying features from a data set according to claim 22, wherein said processor is configured to repeatedly derive new features based upon differently selected segments of interest, the derived features collectively defining a feature set.
- 29. The system for identifying features from a data set according to claim 22, wherein said transformation function is based upon at least one of a weighted mask, a percentile function, and at least one computation derived from at least one subsection of said segment of interest.
- 30. The system for identifying features from a data set according to claim 29, wherein said feature is based upon at least one of a correlation and a covariance computation of said subsection.
- 31. The system for identifying features from a data set according to claim 29, wherein said feature is based upon a distance between at least two subsections of said segments of interest.
- 32. The system for identifying features from a data set according to claim 22, wherein said transformation function is derived by:
breaking said segment of interest into a collection of subsections; and, performing a function that maps at least one subsection to at least one vector.
- 33. The system for identifying features from a data set according to claim 32, wherein said at least one vector is transformed into a signature by clustering said at least one vector across said data set, wherein each data object is characterized in a frequency table indicating how many vectors for that subset are in each cluster.
- 34. The system for identifying features from a data set according to claim 22, wherein said transformation function is derived by:
deconstructing said segments of interest into subsections expressed as 6I=⋃l∈ΛSegl, where I is the segment and Segl is a subsection of said segment of interest; and, letting f: Seg→Rk map a subsection to a vector.
- 35. The system for identifying features from a data set according to claim 22, wherein said data set comprises a collection of images, and said transformation function expands said at least one segment into the pixel gray.
- 36. The system for identifying features from a data set according to claim 22, wherein said transformation function is derived by:
gathering values of primitives; summarizing a distribution of said primitives; and, applying the summarization across at least a portion of said data set.
- 37. A computer system for identifying features from a data set comprising:
a storage device having a plurality of digital representations of data objects stored thereon; an input device configured to accept input from a user; a processor coupled to said storage device and said input device programmed to:
execute a first operation arranged to select a data subset from said plurality of digital representations of data objects; execute a second operation arranged to selectively process said data subset to produce a transformed data subset; execute a third operation arranged to select at least one segment of interest from at least one of said data subset and said transformed data subset; execute a fourth operation arranged to provide at least one tag for said at least one segment of interest; execute a fifth operation arranged to selectively provide at least one characteristic for said at least one segment of interest; execute a sixth operation arranged to construct a transformation function based upon said at least one segment of interest to derive at least one of a feature and a signature from said transformation function; and, execute a seventh operation arranged to make at least one of said feature and said signature available to other processes.
- 38. The computer system for identifying features from a data set according to claim 37, wherein said processor is further programmed to process said data subset prior to deriving said feature to accentuate at least one aspect of interest.
- 39. The computer system for identifying features from a data set according to claim 37, wherein said processor is further programmed to identify a characteristic from said at least one segment of interest, wherein said transformation function is constructed further based further upon said characteristic.
- 40. The computer system for identifying features from a data set according to claim 37, wherein said characteristic comprises identifying two or more segments of interest from the group consisting of similar, distinct, dissimilar, included, excluded, different, identical, mutually exclusive, related, unrelated, and segments that should be ignored.
- 41. The computer system for identifying features from a data set according to claim 37, wherein said processor is further programmed to:
select additional regions of interest; derive at least one of additional features and additional signatures therefrom; and, output at least one of said additional features and additional signatures.
- 42. A computer readable carrier including feature selection program code that causes a computer to perform operations comprising:
executing a first operation arranged to select a data subset from a plurality of digital representations of data objects on a storage medium accessible by said computer readable carrier during execution; executing a second operation arranged to selectively process said data subset to produce a transformed data subset; executing a third operation arranged to select at least one segment of interest from at least one of said data subset and said transformed data subset; executing a fourth operation arranged to provide at least one tag for said at least one segment of interest; executing a fifth operation arranged to selectively provide at least one characteristic for said at least one segment of interest; executing a sixth operation arranged to construct a transformation function based upon said at least one segment of interest to derive at least one of a feature and a signature from said transformation function; and, executing a seventh operation arranged to output at least one of said feature and said signature available to other processes.
- 43. A method of generating a feature from a data set comprising:
evaluating a region of interest of said data set by:
selecting a data subset from said data set; selecting at least one segment of interest from said data subset; and, providing at least one tag for said at least one segment of interest; providing characteristics of said at least one segment of interest; constructing a transformation function based upon said at least one segment of interest; and, deriving a feature from said transformation function.
CROSS REFERENCE TO RELATED APPLICATIONS
[0001] This application claims priority of Provisional application no. 60/275,882 filed Mar. 14, 2001, which is herein incorporated by reference.
Provisional Applications (1)
|
Number |
Date |
Country |
|
60275882 |
Mar 2001 |
US |