METHOD TO DETECT FACES ON IMAGE USING CLASSIFIERS CASCADE

摘要

FIELD: information technologies.;SUBSTANCE: in method to detect faces on image, including application of a cascade of uniform classifiers, each classifier is used to divide the input vector into two classes - "face", "not a face", besides, each classifier is characterised with a higher accuracy of classification compared to a previous classifier in the cascade, circles are sought for on an incoming image using Hough transformation for circles, a set of fragments is formed in image areas, where circles were detected, by division of areas onto a set of fragments with identical size, cascade of classifiers is applied to each fragment, and it is identified whether it contains a face or not, besides, to make a decision on availability of a face all classifiers must refer the appropriate input vector built on the fragment to a "face" class.;EFFECT: increased efficiency and accuracy of processed data as faces are detected on an image.;6 cl, 7 dwg