Operator dependence of methods for obtaining metric characteristics of a face with real measurements and digital images

Abstract

BACKGROUND: No studies in the scientific literature have established the operator dependence of portrait identification with the manual method of measuring facial parameters in subjects compared with measurements from digital images. The labor costs and accuracy of the measurements with these methods have not been established.
 AIM: To establish the operator dependence of methods of real-face measurement and measurement on digital images.
 MATERIALS AND METHODS: Facial parameters were instrumentally measured in 24 Caucasian women aged 19–20 years who were studying at the Professor V.F. Voino-Yasenetsky Krasnoyarsk State Medical University by four researchers independently of each other. The same type of standard digital photography of each subject’s face was taken in five projections, followed by a comparison of the results of real measurements with digital ones.
 RESULTS: When comparing the results of facial measurements obtained from digital images and manual method, an error was observed both in the measurements of one parameter by one researcher using two methods and in the excellent results obtained by other researchers. However, the greatest deviations were observed with the manual method, which may be due to the dependence of measurements on the operator.
 CONCLUSION: With a streamlined and methodically verified approach to working with digital facial images, this technique is more accurate and less labor-intensive than real measurements because of the absence of operator dependence, which can be used in investigating crimes.