Investigating Methods of Input Data Preparation for Person Identification Information Technology

This paper describes the investigation of image preprocessing methods that can be used for preparation of input image data for the algorithm underlying in person identification information technology. Experiments were performed on Python software with developed methods of anisotropic diffusion and image histogram equalization, separately and simultaneously applied, on three sets of images from different databases. As a result, method of image histogram equalization in most experiment cases did not improved the identification accuracy rate of the algorithm. For image from SCface database the identification accuracy rate is stably equal to 92.5%. However, on the images from the Database of Faces the promising result was obtained, that improved algorithm performance on 5%.