Just a while ago, face recognition was being considered to jumped from sci-fi movies. Today though, it keeps expanding its role in our daily life.
Pioneers of Face Recogniton
In 1964, scientists Woodrow Bledsoe, Helen Chan and Charles Bisson started researching about face recognition. These researches as funded by an unanymous intelligence service. That’s why some of the data kept secret.
Huge photo archives-mostly mug shots- were submitted to researchers. And being asked to find the photograph similar to submitted one. If they could find a few close results; it would considered as a victory.
There were a lot of challanges;
* Different images because of different angles of the head
* perspective and distance differences
and so on.
It was impossible to detect pictures of same person because of perspective difference.
This project was called as “human machine” because human features and measurements were being compared. RAND Tablets were being used in this project. It was the state of the art technology back then. This program was measuring the biometrics like irises, eye corners, forehead line and calculating 20 different parameters such as mouth width, eye width, distance between irises etc. Also it could analyse 40 pictures in one hour in this way. Then this data were keeping in data base with the names of photograph owners. In the face recognition phase; these parameters were compared with the others and closest results were given.
2 different phots will have 2 sifferent head angles, leans, and distance from camera. To get this over with, angle, leaning, rotation values must be determined and recalculated for the situation of 0 angle. To make this happen, computer must be capable of calculating the head in 3D environment. Scientist Woodrow Bledsoe calculated a standart that derives these results.
After the resign of Bledsoe in 1966; Peter Hart continued to research and managed to get results with the database of 2000 photographs.
To summerise; despite the limited processor power and resources of that era, biometric face recognition researches were successful.