Image processing apparatus for detecting a face
An electronic image is received by a system to process the image for the presence of a face. The image is repeatedly electronically scanned using a plurality of windows for the presence of facial poses. A plurality of directional poses is detected during the scanning process. Reliabilities for each type of detected poses are calculated. The reliabilities are based on the amount of times the directional poses are detected during the scanning process and directions of the directional poses.
1. A method of using an image processing apparatus, the method comprising: receiving an image at the image processing apparatus; repeatedly scanning the image for a face, the face being in a directional pose; detecting a first pose and a second pose of the face; and determining reliabilities of the first and second pose being the directional pose based on a number of times each pose is detected during repeatedly electronically scanning the electronic image.
2. The method of claim 1 , wherein repeatedly scanning comprises, sequentially setting and moving a detection window over a plurality of positions on the image, and processing the image within the set detection window at each respective position.
3. The method of claim 2 , wherein sequentially setting and moving the detection window is performed over the entirety of the image for a predetermined number of cycles with the detection window reducing in size at the beginning of each cycle.
4. The method of claim 3 , wherein processing a set portion of the image within the set detection window comprises extracting the presence of facial characteristics of the first or second pose using at least one filter.
5. The method of claim 4 , wherein the first pose comprises a first face direction and a tilted first face direction, and wherein the second pose comprises a second face direction and a tilted second face direction.
6. The method of claim 5 , wherein the image is repeatedly scanned in its entirety for the first face direction before the second face direction second, with the tilted face directions following the second face direction.
7. The method of claim 6 , wherein after the image is repeatedly scanned for the second face direction, the image is rotated in a first direction relative to a position of the image when scanned for the first and second face directions and electronically scanned for the tilted first face direction and tilted face profile direction.
8. The method of claim 5 , wherein the image is rotated in a second direction relative to the position of the image when scanned for the first and second face directions and electronically scanned for the tilted first face direction and tilted second face direction.
9. The method of claim 4 , wherein each the first and second face directions are one of a frontal, profile or upward/downward direction.
10. The method of claim 4 , wherein a plurality of facial characteristics of at least one of the poses are respectively detected over a plurality of different cycles within respective set detection windows, and
wherein the set detection windows overlap one another on the image in a predetermined manner, and
wherein the method further comprises unifying the overlapping set detection windows into a single successful detection window.
11. The method of claim 10 , wherein facial characteristics of the first and second poses are respectively detected within first and second successful detection windows, and
wherein the first and second successful detection windows each comprise a set detection window or the successful detection window.
12. The method of claim 11 , further comprising:
determining the first and second successful detection windows overlap one another in an overlap region of the image.
13. The method of claim 12 , wherein determining reliabilities comprises determining that a first reliability is greater than, equal to, or less than a second reliability,
wherein the first and second reliabilities are determined from the amount of one or more set detection windows they each comprise.
14. The method of claim 13 , wherein the first reliability is further determined from a first coefficient, the first coefficient increasing or decreasing the first reliability based on a preferred aspect of the first pose, and
wherein the second reliability is further determined from a second coefficient, the second coefficient increasing or decreasing the second reliability based on a preferred aspect of the second pose.
15. The method of claim 14 , wherein the first and second coefficients are further determined from the source of the image.
16. The method of claim 13 , wherein the first reliability is determined to be equal to the second reliability, and
wherein determining reliabilities further comprises determining that the first pose and second pose have the same or different profile directions, and
wherein one of the different profile directions has a preferred reliability over the other.
17. The method of claim 16 , wherein the profile directions are determined to be the same, and
wherein determining reliability further comprises determining that the first pose and second pose have the same or different profile tilts, and
wherein one of the different profile tilts has a preferred reliability over the other.
18. The method of claim 17 , wherein the profile tilts are determined to be the same which results in a determination that the first and second poses are the same, and
wherein the first and second successful detection windows are unified as a result.
19. A non-transitory computer readable storage medium comprising instructions which when executed by a processing system causes the processing system to perform the method of at least claim 1 .
20. A processing system comprising a processor electronically coupled to the computer readable medium of at least claim 19 .