Methods and systems for extracting feature descriptors for an image
The present invention discloses methods and systems face recognition. Face recognition involves receiving an image/frame, detecting one or more faces in the image, detecting feature points for each of the detected faces in the image, aligning and normalizing the detected feature points, extracting feature descriptors based on the detected feature points and matching the extracted feature descriptors with a set of pre-stored images for face recognition.
1. A feature extraction system with improved accuracy and efficiency in detecting facial feature points comprising of:
a processor, a non-transitory storage element coupled to the processor, encoded instructions stored in the non-transitory storage element, wherein the encoded instructions when implemented by the processor, configure the feature extraction system to:
generate a set of Gabor Magnitude Images (GMIs) for the image using multi-orientation (p) and multi-scale (q) Gabor filters by a filtering unit, wherein the set of GMIs comprises of p*q GMIs, and wherein values for each of the p and q are selected based on at least one of an inter class feature distance and intra class feature distance computed for a set of training images;
generate a set of Local Gabor Binary Pattern (LGBP) maps by computing local binary patterns for each GMI of the set of GMIs using one or more local binary codes by a mapping unit, wherein the one or more local binary codes are quantized into one or more levels, and wherein the set of LGBP maps comprises p*q LGBP maps;
select a plurality of regions in the image by a region selection unit, wherein a region of the plurality of regions is determined by a feature point of a set of feature points detected in the image, wherein the set of feature points is detected using a deformable part-based model (CDPM), wherein the CDPM is trained using a latent Support Vector Machine (SVM), wherein at least one of geometry and location of parts form a set of latent variables, and wherein the feature extraction system further comprises a new feature generation unit configured to generate one or more new facial feature points of the image based on at least one of the set of facial feature points detected using the CDPM; and
generate p*q histograms for each region of the plurality of regions by a feature description unit, wherein a histogram of the p*q histograms is generated for a LGBP map of the p*q LGBP maps;
whereby histograms thus generated for the plurality of regions are the feature descriptors for the image, enabling improved accuracy and efficiency in detecting facial feature points of the images.
2. The feature extraction system of claim 1 , wherein the image is a face.
3. The feature extraction system of claim 2 , wherein the image is at least one of the group comprising an intensity map of the face, individual color channels of the face and a transformed color space of the face.
4. The feature extraction system of claim 1 , wherein the image corresponds to a frame of a video.
5. The feature extraction system of claim 1 , wherein the values for p range from 5 to 15.
6. The feature extraction system of claim 1 , wherein the values for q range from 3 to 12.
7. The feature extraction system of claim 1 further comprising a post processor configured to transform the histograms to a lower dimensional subspace.
8. A computer-implemented method for extracting feature descriptors for an image using a processor, a non-transitory storage element coupled to the processor and encoded instructions stored in the non-transitory storage element, the method comprising:
generating a set of Gabor Magnitude Images (GMIs) for the image using multi-orientation (p) and multi-scale (q) Gabor filters, wherein the set of GMIs comprises of p*q GMIs, and wherein values for each of the p and q are selected based on at least one of an inter class feature distance and intra class feature distance computed for a set of training images;
generating a set of Local Gabor Binary Pattern (LGBP) maps by computing local binary patterns for each GMI of the set of GMIs using one or more local binary codes, wherein the one or more local binary codes are quantized into one or more levels, and wherein the set of LGBP maps comprises p*q LGBP maps;
selecting a plurality of regions in the image, wherein a region of the plurality of regions is determined by a feature point of a set of feature points detected in the image, wherein the set of feature points is detected using a deformable part-based model (CDPM), wherein the CDPM is trained using a latent Support Vector Machine (SVM), wherein at least one of geometry and location of parts form a set of latent variables, and wherein the feature extraction system further comprises a new feature generation unit configured to generate one or more new facial feature points of the image based on at least one of the set of facial feature points detected using the CDPM; and
generating p*q histograms for each region of the plurality of regions, wherein a histogram of the p*q histograms is generated for a LGBP map of the p*q LGBP maps;
whereby histograms thus generated for the plurality of regions are the feature descriptors for the image, enabling accurate and efficient recognition of the images.
9. The method for extracting feature descriptors of claim 8 , wherein the image is a face.
10. The method for extracting feature descriptors of claim 9 , wherein the image is at least one of the group comprising an intensity map of the face, individual color channels of the face and a transformed color space of the face.
11. The method for extracting feature descriptors of claim 8 further comprising transforming the histograms to a lower dimensional subspace.
12. A feature extraction system comprising of:
a processor, a non-transitory storage element coupled to the processor, encoded instructions stored in the non-transitory storage element, wherein the encoded instructions when implemented by the processor, configure the feature extraction system to:
generate a set of Gabor Magnitude Images (GMIs) for the image using multi-orientation (p) and multi-scale (q) Gabor filters by a filtering unit, wherein the set of GMIs comprises of p*q GMIs, and wherein values for each of the p and q are selected based on at least one of an inter class feature distance and intra class feature distance computed for a set of training images;
generate a set of Local Gabor Binary Pattern (LGBP) maps by computing local binary patterns for each GMI of the set of GMIs using one or more local binary codes by a mapping unit, wherein the one or more local binary codes are quantized into one or more levels, and wherein the set of LGBP maps comprises p*q LGBP maps;
select a plurality of regions in the image, wherein a region of the plurality of regions is determined by a facial feature point of a set of facial feature points detected in the image, wherein the set of feature points is detected using a deformable part-based model (CDPM), wherein the CDPM is trained using a latent Support Vector Machine (SVM), wherein at least one of geometry and location of parts form a set of latent variables, and wherein the feature extraction system further generates one or more new facial feature points of the image based on at least one of the set of facial feature points detected using the CDPM; and
generate p*q histograms for each region of the plurality of regions, wherein a histogram of the p*q histograms is generated for each LGBP map of the p*q LGBP maps;
whereby histograms thus generated for the plurality of regions are the feature descriptors for the image, enabling accurate and efficient recognition of the images.
13. The feature extraction system of claim 12 , wherein the image is at least one of the group comprising an intensity map, individual color channels and a transformed color space of the image.
14. The feature extraction system of claim 12 comprising a post processor configured to transform the histograms to a lower dimensional subspace.