Method, apparatus and computer program product for human-face features extraction

1. A method comprising:

detecting a same human-face image in a plurality of image frames of a video sequence;

dividing the detected human-face image into a plurality of local patches with a predetermined size, wherein each local patch is around or near a human-face feature point;

determining a correspondence relationship between respective local patches of the same human-face image in the plurality of image frames of the video sequence;

determining a first local feature from a first frame of the plurality of image frames of the same human-face image;

determining a second local feature, different than the first local feature, from a second frame of the plurality of image frames of the same human-face image;

using human-face local feature vector components extracted from respective local patches having a mutual correspondence relationship based on the first local feature and the second local feature of the first and the second frames of the same human-face image to form human-face local feature vectors representing the facial points corresponding to the local patches, wherein formation of the human-face local feature vectors comprises resizing the plurality of local patches to obtain the human-face local feature vectors representing the facial points corresponding to respective resized local patches;

and

combining the resulting human-face local feature vectors representing the facial points corresponding to respective local patches to form human-face global feature vectors.

2. The method of claim 1 , wherein using human-face local feature vector components extracted from respective local patches having a mutual correspondence relationship to form the human-face local feature vectors representing the facial points corresponding to the local patches comprises:

identifying different poses of the same human-face image in a plurality of image frames of a video sequence;

extracting human-face local feature vector components only from respective un-occluded local patches having a mutual correspondence relationship based on the identified different poses of the same human face in the plurality of image frames; and

combining the human-face local feature vector components extracted from respective un-occluded local patches having a mutual correspondence relationship to form a human-face local feature vectors representing the facial points corresponding to the local patches.

3. The method of claim 1 , wherein using human-face local feature vector components extracted from respective local patches having a mutual correspondence relationship to form human-face local feature vectors representing the facial points corresponding to the local patches comprises:

identifying different poses of the same human-face image in a plurality of image frames of a video sequence; and

weight combining the human-face local feature vector components extracted from respective local patches having a mutual correspondence relationship based on the identifed different poses of the same human face in different image frames to form human-face local feature vectors representing the facial points corresponding to the local patches.

4. The method of claim 1 , wherein dividing the detected human-face image into a plurality of local patches with a predetermined size is implemented by adding a grid on the detected human-face image.

5. The method of claim 1 , further comprising combining the resulting human-face local feature vectors representing the facial points corresponding to respective resized local patches to form human-face global feature vectors.

6. The method of claim 5 , further comprising combining human-face global feature vectors obtained under different local patch sizes to form a human-face global feature vector set.

7. The method of claim 1 , further comprising performing human-face recognition using the resulting human-face global feature vectors.

8. An apparatus, comprising:

at least one processor and at least one memory including a computer program code, wherein the at least one memory including the computer program code are configured, with the at least one processor, to cause the apparatus to:

detect a same human-face image in a plurality of image frames of a video sequence;

divide the detected human-face image into a plurality of local patches with a predetermined size, wherein each local patch is around or near a human-face feature point;

determine a correspondence relationship between respective local patches of the same human-face image in the plurality of image frames of the video sequence;

determine a first local feature from a first frame of the plurality of image frames of the same human-face image;

determine a second local feature, different than the first local feature, from a second frame of the plurality of image frames of the same human-face image;

use human-face local feature vector components extracted from respective local patches having a mutual correspondence relationship based on the first local feature and the second local feature of the first and the second frames of the same human-face image to form human-face local feature vectors representing facial points corresponding to the local patch, wherein formation of the human-face local feature vectors comprises resizing the plurality of local patches to obtain the human-face local feature vectors representing the facial points corresponding to respective resized local patches;

and

combine the resulting human-face local feature vectors representing the facial points corresponding to respective local patches to form human-face global feature vectors.

9. The apparatus of claim 8 , wherein dividing the detected human-face into a plurality of local patches with a predetermined size is implemented by adding a grid on the detected human-face image.

10. The apparatus of claim 8 , wherein the at least one memory including the computer program code are configured, with the at least one processor, to further cause the apparatus to combine the resulting human-face local feature vectors representing the facial points corresponding to respective resized local patches to form a human-face global feature vector.

11. The apparatus of claim 10 , wherein the at least one memory including the computer program code are configured, with the at least one processor, to further cause the apparatus to combine the resulting human-face global feature vectors under different local patch sizes to form a human-face global feature vector set.

12. The apparatus of claim 8 , wherein the at least one memory including the computer program code are configured, with the at least one processor, to further cause the apparatus to use the resulting human-face global feature vectors to perform human-face recognition.

13. The apparatus of claim 11 , wherein the at least one memory including the computer program code are configured, with the at least one processor, to further cause the apparatus to use the resulting human-face global feature vector set to perform human-face recognition.

14. An apparatus comprising:

at least one processor and at least one memory including a computer program code, wherein the at least one memory including the computer program code are configured, with the at least one processor, to cause the apparatus to:

detect a same human-face image in a plurality of image frames of a video sequence;

divide the detected human-face image into a plurality of local patches with a predetermined size, wherein each local patch is around or near a human-face feature point;

determine a correspondence relationship between respective local patches of the same human-face image in the plurality of image frames of the video sequence;

determine a first local feature from a first frame of the plurality of image frames of the same human-face image;

determine a second local feature, different than the first local feature, from a second frame of the plurality of image frames of the same human-face image; and

use human-face local feature vector components extracted from respective local patches having a mutual correspondence relationship based on the first local feature and the second local feature of the first and the second frames of the same human-face image to form human-face local feature vectors representing facial points corresponding to the local patch, wherein the apparatus is further caused to:

identify different poses of the same human-face image in a plurality of image frames of a video sequence;

extract the human-face local feature vector components only from respective un-occluded local patches having a mutual correspondence relationship based on the identified different poses of the same human face in the plurality of image frames; and

combine the human-face local feature vector components extracted from respective un-occluded local patches having a mutual correspondence relationship to form the human-face local feature vectors representing facial points corresponding to the local patches.

15. An apparatus

at least one processor and at least one memory including a computer program code, wherein the at least one memory including the computer program code are configured, with the at least one processor, to cause the apparatus to:

detect a same human-face image in a plurality of image frames of a video sequence;

divide the detected human-face image into a plurality of local patches with a predetermined size, wherein each local patch is around or near a human-face feature point;

determine a correspondence relationship between respective local patches of the same human-face image in the plurality of image frames of the video sequence;

determine a first local feature from a first frame of the plurality of image frames of the same human-face image;

determine a second local feature, different than the first local feature, from a second frame of the plurality of image frames of the same human-face image; and

use human-face local feature vector components extracted from respective local patches having a mutual correspondence relationship based on the first local feature and the second local feature of the first and the second frames of the same human-face image to form human-face local feature vectors representing facial points corresponding to the local patch, wherein the apparatus is further caused to:

identify different poses of the same human-face image in a plurality of image frames of a video sequence; and

weight combine the human-face local feature vector components extracted from respective local patches having a mutual correspondence relationship based on the identified different poses of the same human face in different image frames to form human-face local feature vectors representing the facial points corresponding to the local patches.