|
Sci Tech
Rating a smile using face recognition software
How many times have you missed capturing the best smile and had to settle for a ‘not–so smile?’ If only you had an ‘intelligent camera’, that will ‘click’ when the ‘smile is 100 per cent,’ you would have saved time and agony.
Well, Omron Corporation ( www.omron.com) has announced a ‘Smile Measurement Software’ that can rate the smile on a scale of 0 -100 per cent and demonstrated it at the 2007 Combined Exhibition of Advanced Technologies (CEATEC), Japan.
When one smiles, the eyes and mouth open and their shapes change. The software uses 3D face mapping technology that detects and analyzes these features to measure the smile accurately. A Pentium PC with 3.2 GHZ clock speed takes only 0.044 second to process.
However, the faces must at least be sixty pixels wide and tilted less than thirty degrees either side and fifteen degrees up and down.
It was developed after studying the faces of 15,000 people (babies to elderly) from different cultures. Its size (46 kb) makes it ideal for mobile devices.
Face Recognition technology has been in use for quite some time. It involves determining key points or nodal points (such as tip of the nose, distance between eyes, depth of eye sockets, and length of jaw line) from a photograph and comparing them with those of photos in a database for match.
While there are about 80 nodal points, about 20 are sufficient for recognition. The golden triangle refers to the face from temple to temple and just above the lip. The features inside this triangle are unaffected by hirsuteness, wearing of spectacles, and ageing.
Omron claims that it can “detect and analyze faces even in blurred or partially obscured images or when the subject is not looking directly at the camera” and it can also determine gender, age and ethnicity with facial image.
3D recognition involves the use of 3D images or video to record the faces under different lighting conditions and viewing angles. These tend to be more accurate than 2D recognition. Mathematical algorithms facilitate the production of 2D images from 3D images and also vice versa.
A great development is the use of microfeature analysis — identification of patterns of pores and grains in skin texture.
Combining face and skin texture, the accuracy of recognition improves significantly. Possible changes due to lighting, angle of photo, blinking etc are also accounted for.
Technology exists to even identify faces in a crowd and from other surrounding objects. Face Recognition technology is primarily used for identification, authentication and authorization for using computers, ATMs and entering buildings.
The US-VISIT program of the U.S. government involves recording of the fingerprint and photo when a visa is given and comparing those with the actual ones taken at the time of entry into U.S. to ascertain that it is the same person.
S. CHELLIAH
(The author is a Consultant with Satyam Computer Services Ltd. Chellaiah_s@satyam.com)
Printer friendly
page
Send this article to Friends by
E-Mail
Sci Tech
|
