A method and system for image recognition in a collection of digital images includes training image classifiers and retrieving a sub-set of images from the collection. For each image in the collection, any regions within the image that correspond to a face are identified. For each face region and an
A method and system for image recognition in a collection of digital images includes training image classifiers and retrieving a sub-set of images from the collection. For each image in the collection, any regions within the image that correspond to a face are identified. For each face region and any associated peripheral region, feature vectors are determined for each of the image classifiers. The feature vectors are stored in association with data relating to the associated face region. At least one reference region including a face to be recognized is/are selected from an image. At least one classifier on which said retrieval is to be based is/are selected from the image classifiers. A respective feature vector for each selected classifier is determined for the reference region. The sub-set of images is retrieved from within the image collection in accordance with the distance between the feature vectors determined for the reference region and the feature vectors for face regions of the image collection.
대표청구항▼
The invention claimed is: 1. A method for image recognition in a collection of digital images comprising: using a processor to perform the following steps: a) training a plurality of image classifiers, including: for each image in the collection, identifying any regions in the image corresponding
The invention claimed is: 1. A method for image recognition in a collection of digital images comprising: using a processor to perform the following steps: a) training a plurality of image classifiers, including: for each image in the collection, identifying any regions in the image corresponding to a face; for each face region and any associated peripheral region, for each of a plurality of image classifiers, determining feature vectors; and storing said feature vectors in association with data relating to the associated face region, and b) retrieving a sub-set of images from said collection, including: selecting from an image at least one reference region including a face to be recognized; selecting from said plurality of image classifiers, at least one classifier on which said retrieval is to be based; determining for said reference region, a respective feature vector for each selected classifier; and retrieving said sub-set of images from within said image collection in accordance with the distance between the feature vectors determined for said reference region and the feature vectors for face regions of said image collection wherein said step of determining feature vectors comprises: a. for each face region and any associated peripheral region, extracting respective features representative of the region; b. for each of said plurality of image classifiers, determining respective basis vectors according to said extracted features; and c. for the extracted features for each region, for each classifier, determining said feature vectors, based on each determined basis vector. 2. A method as claimed in claim 1 wherein said training step comprises: normalizing identified face regions and associated peripheral regions. 3. A method as claimed in claim 1 wherein said peripheral regions are indicative of hair, clothing and pose characteristics of a person associated with said identified face region. 4. A method as claimed in claim 1 wherein each basis vector for a classifier comprises a selected set of attributes and respective reference values for these attributes. 5. A method as claimed in claim 4 wherein the reference value for an attribute comprises an origin value. 6. A method as claimed in claim 4 wherein the reference value for an attribute comprises a mean value across all identified regions. 7. A method as claimed in claim 4 wherein each feature vector comprises an identified region's attribute values within the basis vector space relative to the reference value. 8. A method as claimed in claim 1 wherein said step of selecting a reference region comprises a user selecting a face region and any peripheral region from a reference image. 9. A method as claimed in claim 1 wherein said step of selecting a reference region comprises automatically identifying and selecting one or more face regions and any peripheral regions from a reference image. 10. A method as claimed in claim 1 wherein said step of determining feature vectors for said reference region, is responsive to determining that feature vectors have previously been determined for said reference region for said classifier, for retrieving said feature vectors from storage. 11. A method as claimed in claim 1 wherein retrieving said sub-set of images comprises, for each classifier, comparing feature vectors for the selected face region with feature vectors for face regions in the image collection to provide a set of distance measures. 12. A method as claimed in claim 1 wherein classifiers for face regions include a wavelet domain PCA (principle component analysis) classifier; and a 2-dimensional DCT (discrete cosine transform) classifier. 13. A method as claimed in claim 12 comprising filtering high frequency components from an identified face region prior to determining the feature vectors for said wavelet domain PCA classifier. 14. A method as claimed in claim 12 wherein determining a feature vector for a face region for said 2-D DGT classifier includes: (i) transforming image information for an identified normalized face region to an indexed image using a 256 color gif colormap; and (ii) applying a 2D DCT transform to said transformed image information to provide a spectrum for use as a feature vector for said face region. 15. A portable digital camera including a lens, an image sensor and a processor, and having an operating system embodied within a processor-readable medium for programming the processor to perform a method as claimed in claim 1. 16. A method for image recognition in a collection of digital images comprising: using a processor to perform the following steps: a) training a plurality of image classifiers, including: for each image in the collection, identifying any regions in the image corresponding to a face; for each face region and any associated peripheral region, for each of a plurality of image classifiers, determining feature vectors; and storing said feature vectors in association with data relating to the associated face region, and b) retrieving a sub-set of images from said collection, including: selecting from an image at least one reference region including a face to be recognized; selecting from said plurality of image classifiers, at least one classifier on which said retrieval is to be based; determining for said reference region, a respective feature vector for each selected classifier; and retrieving said sub-set of images from within said image collection in accordance with the distance between the feature vectors determined for said reference region and the feature vectors for face regions of said image collection, and wherein retrieving said sub-set of images comprises, for each classifier, comparing feature vectors for the selected face region with feature vectors for face regions in the image collection to provide a set of distance measures; and wherein the method further comprising calculating for each set of distance measures, mean and variance values. 17. A method as claimed in claim 16 comprising normalizing said set of distance measures using said mean and variance values. 18. A method as claimed in claim 17 comprising summing said normalized sets of distance measures for each selected classifier to provide a combined distance measure from said reference region to each region in said image collection. 19. A method as claimed in claim 18 comprising ranking each image in said image collection according to said combined distance measures and selecting said sub-set of images according to said ranking. 20. A method for image recognition in a collection of digital images comprising: using a processor to perform the following steps: a) training a plurality of image classifiers, including: for each image in the collection, identifying any regions in the image corresponding to a face; for each face region and any associated peripheral region, for each of a plurality of image classifiers, determining feature vectors; and storing said feature vectors in association with data relating to the associated face region, and b) retrieving a sub-set of images from said collection, including: selecting from an image at least one reference region including a Thee to be recognized; selecting from said plurality of image classifiers, at least one classifier on which said retrieval is to be based; determining for said reference region, a respective feature vector for each selected classifier; and retrieving said sub-set of images from within said image collection in accordance with the distance between the feature vectors determined for said reference region and the feature vectors for face regions of said image collection, and wherein classifiers for face regions include a wavelet domain PCA (principle component analysis) classifier; and a 2-dimensional DCT (discrete cosine transform) classifier, and wherein determining a feature vector for a face region for said wavelet domain PCA classifier includes: (i) transforming image information for an identified normalized face region into a gray scale; (ii) applying wavelet decomposition to said transformed image information to provide a sub-band representation of the face region; (iii) performing histogram equalization on the sub-band representation; and (iv) subtracting a mean sub-band value from the equalized sub-band representation. 21. A method for image recognition in a collection of digital images comprising: using a processor to perform the following steps: a) training a plurality of image classifiers, including: for each image in the collection, identifying any regions in the image corresponding to a face; for each face region and any associated peripheral region, for each of a plurality of image classifiers determining feature vectors; and storing said feature vectors in association with data relating to the associated face region, and b) retrieving a sub-set of images from said collection, including: selecting from an image at least one reference region including a face to be recognized; selecting from said plurality of image classifiers, at least one classifier on which said retrieval is to be based; determining for said reference region, a respective feature vector for each selected classifier; and retrieving said sub-set of images from within said image collection in accordance with the distance between the feature vectors determined for said reference region and the feature vectors for face regions of said image collection, and wherein respective classifiers for peripheral regions are based on color histogram, color moment, colour correlogram, banded colour correlogram, and wavelet texture analysis. 22. A component embodied within a computer-readable medium for programming a processor to perform a method including image recognition in a collection of digital images, the component comprising: a) training module for training a plurality of image classifiers, said module being arranged: for each image in the collection, to identify any regions in the image corresponding to a face; for each face region and any associated peripheral region, for each of a plurality of image classifiers, to determine feature vectors; and to store said feature vectors in association with data relating to the associated face region, and b) a retrieval module for retrieving a sub-set of images from said collection and arranged: to select from an image at least one reference region including a face to be recognized; to select from said plurality of image classifiers, at least one classifier on which said retrieval is to be based; to determine for said reference region, a respective feature vector for each selected classifier; and to retrieve said sub-set of images from within said image collection in accordance with the distance between the feature vectors determined for said reference region and the feature vectors for face regions of said image collection, and said image browser comprising a user interface arranged to enable a user to define said image collection by selecting one or more folders from a hierarchy of folders, each of said one or more folders storing one or more images, and p1 wherein said training module is responsive to said user defined image collection containing images for two or more previously trained image collections for combining the basis vectors for said image collections. 23. A component as claimed in claim 22 in which said training module is arranged to identify any regions in the image corresponding to a face only once. 24. An image browser including a component as claimed in claim 22, said image browser comprising a user interface arranged to enable a user to define said image collection by selecting one or more folders from a hierarchy of folders, each of said one or more folders storing one or more images. 25. An image browser according to claim 22 wherein said training module is responsive to a user adding less than a threshold number of images to a previously trained image collection for determining feature vectors for face regions of said added images relative to existing basis vectors for each classifier. 26. An image browser according to claim 22, said user interface arranged to enable a user to define said image collection by selecting one or more images from one or more folders. 27. A component according to claim 22 comprising a user interface including a respective user interface device associated with each classifier and wherein said at least one classifier is user selected through selecting a corresponding at least one of said user interface devices. 28. An image acquisition device including a processor, and code embodied within a processor-readable medium for programming the processor to perform an image recognition method, including a component as claimed in claim 22. 29. A component embodied within a computer-readable medium for programming a processor to perform a method including image recognition in a collection of digital images, the component comprising: a training module for training a plurality of image classifiers, said module being arranged: for each image in the collection, to identify any regions in the image corresponding to a face; for each face region and any associated peripheral region, for each of a plurality of image classifiers, to determine feature vectors; and to store said feature vectors in association with data relating to the associated face region, and b) a retrieval module for retrieving a sub-set of images from said collection and arranged: to select from an image at least one reference region including a face to be recognized; to select from said plurality of image classifiers, at least one classifier on which said retrieval is to be based; to determine for said reference region, a respective feature vector for each selected classifier; and to retrieve said sub-set of images from within said image collection in accordance with the distance between the feature vectors determined for said reference region and the feature vectors for face regions of said image collection, and said image browser comprising a user interface arranged to enable a user to define said image collection by selecting one or more folders from a hierarchy of folders, each of said one or more folders storing one or more images, and wherein said training module is responsive to a user adding more than a threshold number of images to a previously trained image collection for re-calculating any existing basis vector for each classifier and determining feature vectors for all face regions of said image collection relative to said re-calculated basis vectors. 30. A component embodied within a computer-readable medium for programming a processor to perform a method including image recognition in a collection of digital images, the component comprising: a training module for training a plurality of image classifiers, said module being arranged: for each image in the collection, to identify any regions in the image corresponding to a face; for each face region and any associated peripheral region, for each of a plurality of image classifiers, to determine feature vectors; and to store said feature vectors in association with data relating to the associated face region, and b) a retrieval module for retrieving a sub-set of images from said collection and arranged: to select from an image at least one reference region including a face to be recognized; to select from said plurality of image classifiers, at least one classifier on which said retrieval is to be based; to determine for said reference region, a respective feature vector for each selected classifier; and to retrieve said sub-set of images from within said image collection in accordance with the distance between the feature vectors determined for said reference region and the feature vectors for face regions of said image collection, and further comprising a user interface including a respective user interface device associated with each classifier and wherein said at least one classifier is user selected through selecting a corresponding at least one of said user interface devices, and wherein each said user interface device includes a sub-component arranged to enable a user to set a weighting for each classifier. 31. A component according to claim 30 wherein said retrieval module is arranged to retrieve said sub-set of images from within said image collection for said reference region in accordance with a weighted distance between the feature vectors determined for said reference region and the feature vectors for face regions of said image collection.
연구과제 타임라인
LOADING...
LOADING...
LOADING...
LOADING...
LOADING...
이 특허에 인용된 특허 (143)
Pace,Charles Paul, Apparatus and method for processing video data.
Benati Paul J. (Webster NY) Gray Robert T. (Rochester NY) Cosgrove Patrick A. (Honeoye Falls NY), Automated detection and correction of eye color defects due to flash illumination.
Matsuo, Hideaki; Imagawa, Kazuyuki; Takata, Yuji; Baba, Naruatsu; Ejima, Toshiaki, Device and method for face image extraction, and recording medium having recorded program for the method.
Buhr John D. ; Goodwin Robert M. ; Koeng Frederick R. ; Rivera Jose E., Digital photofinishing system including scene balance, contrast normalization, and image sharpening digital image processing.
Parulski Kenneth A. (Rochester NY) Napoli Thomas A. (Rochester NY) Lewis David M. (Waterport NY), Electronic still camera for capturing and categorizing images.
Kuchta Daniel W. (Brockport NY) Sucy Peter J. (Hamlin NY), Electronic still camera providing multi-format storage of full and reduced resolution images.
Steffens Johannes Bernhard ; Elagin Egor Valerievich ; Nocera Luciano Pasquale Agostino ; Maurer Thomas ; Neven Hartmut, Face recognition from video images.
Prasad K. Venkatesh (Cupertino CA) Stork David G. (Stanford CA), Facial feature extraction method and apparatus for a neural network acoustic and visual speech recognition system.
Poggio Tomaso ; Beymer David ; Jones Michael ; Vetter Thomas,DEX, Image compression by pointwise prototype correspondence using shape and texture information.
Yoichi Takaragi JP; Masanori Yamada JP; Yoshinobu Sato JP; Yasumichi Suzuki JP; Yasuhiro Yamada JP; Akiko Kanno JP; Yoshiki Uchida JP, Image processing apparatus and method for discriminating an original having a predetermined pattern.
Brogliatti, Barbara Spencer; Grakal, Christopher; Janney, Lisa A.; O'Neil, Marisa B.; Smith, Thomas G., Method and apparatus for archiving in and retrieving images from a digital image library.
Tsubouchi Yuzo,JPX ; Abe Nobutaka,JPX ; Nakamura Yuji,JPX, Method and apparatus for mounting parts with pick-up mistake detection and storage of mistake rate.
Zanzucchi,Peter John; Moroney, III,Richard Morgan; Aceti,John Gregory; Pletcher,Timothy Allen; Burstyn,Herschel Clement, Method and imager for detecting the location of objects.
Eckes,Christian; Kefalea,Efthimia; Von Der Malsburg,Chrstoph; P철tzsch,Michael; Rinne,Michael; Triesch,Jochen; Vorbr체ggen,Jan C., Method for recognizing objects in digitized images.
Smith Joseph J. (2601 Knollwood Rd. Charlotte NC 28211) Moseley Thomas L. (9352 Pinewood St. Charlotte NC 28214), Portable electrolytic testing device for metals.
Ogrinc Michael A. (San Francisco CA) Card Robert A. (Palo Alto CA) Burns Chris R. (Mountain View CA) Clarke Charles P. (Los Altos CA) Collier Ronda L. (Scotts Valley CA) Collins Kevin M. (San Mateo C, Real time video image processing system.
Okubo,Atsushi; Sabe,Kohtaro; Kawamoto,Kenta; Fukuchi,Masaki, Robot device and face identifying method, and image identifying device and image identifying method.
Mashimo Yukio (Tokyo JA) Sakurada Nobuaki (Kanagawa JA) Ito Tadashi (Kanagawa JA) Ito Fumio (Kanagawa JA) Shinoda Nobuhiko (Tokyo JA), System for exposure measurement and/or focus detection by means of image senser.
Lee,Kyunghee; Chung,Yongwha; Park,Chee Hang; Byun,Hyeran, System for registering and authenticating human face using support vector machines and method thereof.
Wheeler Richard B. (Webster NY), Technique suited for use in multi-zone autofocusing cameras for improving image quality for non-standard display sizes a.
Kojima Kazuaki (Nagaokakyo JPX) Kuno Tetsuya (Nagaokakyo JPX) Sugiura Hiroaki (Nagaokakyo JPX) Yamada Takeshi (Nagaokakyo JPX), Video signal processor for detecting flesh tones in am image.
Jain Ramesh ; Horowitz Bradley ; Fuller Charles E. ; Gupta Amarnath ; Bach Jeffrey R. ; Shu Chiao-fe, Visual image database search engine which allows for different schema.
Steinberg, Eran; Corcoran, Peter; Prilutsky, Yury; Bigioi, Petronel; Ciuc, Mihai; Ciurel, Stefanita; Verlan, Constantin, Classification and organization of consumer digital images using workflow, and face detection and recognition.
Steinberg, Eran; Corcoran, Peter; Prilutsky, Yury; Bigioi, Petronel; Ciuc, Mihai; Ciurel, Stefanita; Vertan, Constantin, Classification and organization of consumer digital images using workflow, and face detection and recognition.
Steinberg, Eran; Corcoran, Peter; Prilutsky, Yury; Bigioi, Petronel; Ciuc, Mihai; Ciurel, Stefanita; Vertan, Constantin, Classification and organization of consumer digital images using workflow, and face detection and recognition.
Kottke, Dane P.; Guo, John J.; Lee, Jeyun; Park, Sangseok; Weed, Christopher; Kwan, Justin; Lee, Nigel, Continuous block tracking for temporal prediction in video encoding.
Corcoran, Peter; Bigioi, Petronel; Stec, Piotr, Face and other object detection and tracking in off-center peripheral regions for nonlinear lens geometries.
Steinberg, Eran; Bigioi, Petronel; Corcoran, Peter; Gangea, Mihnea; Petrescu, Stefan Mirel; Vasiliu, Andrei; Costache, Gabriel; Drimbarean, Alexandru, Face searching and detection in a digital image acquisition device.
Steinberg, Eran; Bigioi, Petronel; Corcoran, Peter; Gangea, Mihnea; Petrescu, Stefan Mirel; Vasiliu, Andrei; Costache, Gabriel; Drimbarean, Alexandru, Face searching and detection in a digital image acquisition device.
Steinberg, Eran; Bigioi, Petronel; Corcoran, Peter; Gangea, Mihnea; Petrescu, Stefan Mirel; Vasiliu, Andrei; Costache, Gabriel; Drimbarean, Alexandru, Face searching and detection in a digital image acquisition device.
Kyyko, Volodymyr; Matsello, Vyacheslav; Tsariov, Andriy; Musatenko, Yuriy; Schlezinger, Mykhailo; Kyiko, Kostiantyn; Nedashkivskyi, Oleksii, Method and system of person identification by facial image.
Lin, Zhe; Guggisberg, Stefan; Brandt, Jonathan; Marth, Michael, Training a classifier algorithm used for automatically generating tags to be applied to images.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.