Masliy R. Information technology detecting and tracking faces in surveillance tasks

Object of research - processes of face detection and tracking in surveillance tasks; the purpose of work is to increase the likelihood of face detection and tracking in the image in surveillance tasks. Methods of research are based on machine learning methods for the stage of training in detection methods, simulation modeling for creation of tracking method, digital processing of images with using Matlab environment and OpenCV library for pre-processing of images, object-oriented programming in Microsoft Visual Studio Express 2010 for implementation of information technology in the form of software; theoretical results: a method of detection of frontal faces in grayscale images, which uses modified haar-like features in simple classifiers is created for the first time. This increased the likelihood of face detection in images with complex lighting; a method of detection of frontal faces with slopes from the central axis of symmetry in the range of ± 56 degrees in grayscale images which uses rectangular local binary patterns in simple classifiers is developed for the first time. This increased the likelihood of face detection; method of color image segmentation on “skin” and “non-skin” pixels with using of skin color models obtains further development, unlike existing ones, filtering function parameters are denoted because of increasing of the probability of "non-skin" pixels rejecting, thus improving the speed of color image processing and reduce the number of false detections subject to a fixed probability of face detection using the developed methods; method of face tracking in video sequences based on Monte Carlo method and color approach of observation model representation is improved. Its features are using of detection method for initial denotation of face location and updates of face position during the tracking process, thus improving the likelihood of face tracking.The practical significance of the obtained results is that on the basis of new and improved existing methods, a new information technology of face detection and tracking that allowed to increase the probability of detection and tracking under complex lighting and occlusions is created. The proposed information technology can be used in surveillance systems for search task solving in video archives, intelligent video compression, human tracking with his/her face image, person verification with his/her face image. The degree of implementation is the main results are introduced in the private enterprise “TVA-service”, in private enterprise “Anklus” in Vinnitsa Chamber of Commerce and in the studying process of Vinnytsia National Technical University. Sector (industry) of use is Face Recognition and Surveillance Systems.