Age Invariant Face Recognition
Aman Jain &
Nikhil Rasiwasia
Under the Guidance of Prof R M
K Sinha
EE372 - Computer Vision and
Document Processing
· A database of 6 adult
and 6 child images was used.
· For a match of an adult image with all the child images in the database, 4 out of 6 images could be matched
successfully.
· For a match of a child image with all the adult images in the database,again 4 out of 6 images could be
matched successfully.
· There was no usefulness of the distance between eyes for face recognition.
· Most useful feature (definitely a RAI feature ) seems to be the distance between the lips and nose.
· Another important feature is the distance between the lips and eyes.
· In our database, their was a mismatch for pairs number 2 and 4. Rest all matched in both the directions.
· Even though the database was very small, there seems to be a very good possibility of a successful ageinvariant
matching technique to be possible using feature analysis.
· For a match of an adult image with all the child images in the database, 4 out of 6 images could be matched
successfully.
· For a match of a child image with all the adult images in the database,again 4 out of 6 images could be
matched successfully.
· There was no usefulness of the distance between eyes for face recognition.
· Most useful feature (definitely a RAI feature ) seems to be the distance between the lips and nose.
· Another important feature is the distance between the lips and eyes.
· In our database, their was a mismatch for pairs number 2 and 4. Rest all matched in both the directions.
· Even though the database was very small, there seems to be a very good possibility of a successful ageinvariant
matching technique to be possible using feature analysis.
