Search academic texts

Information × Registration Number 0210U007915, 0109U004834 , R & D reports Title Biometric identification of the person in video observation systems popup.stage_title Head Sachenko Anatoliy, Registration Date 05-01-2011 Organization Ternopol National Economic University popup.description2 Research object: video surveillance system. Subject of research: methods for motion detection, face detection, tracking and recognition in video flow. Goal: development of human biometric identification methods using face image for video surveillance systems in order to improve validity and speed with limited computational resources. Research methods: System analysis methods, image processing methods and the theory of artificial neural net-works, mathematical statistics, object-oriented modeling and programming methods. Results and their novelty: the motion detection method is improved for detecting moving objects in video flow in different lighting conditions using dynamic background update coefficient. it allowed increasing validity for foreground/background pixels classi-fication; the video flow image representation method is improved using quadratic tree and integral image. it allowed increasing speed for each frame processing; the generalized face detection information model is developed based on multi-level combined cascade of classifiers. it allowed improving face detection methods for grayscale and color im-ages; the face detection method is developed using the combined cascade of neural network classifiers that includes an additional level for face-candidates verification presented by convolutional neural network. it allowed increasing detection validity; the face-candidate verification method is improved based on face-candidate image processing by convolutional neural network at once. it allowed increasing speed of the whole combined cascade of neural network classifiers. Product Description popup.authors Довгань Володимир В'ячеславович Палій Ігор Орестович popup.nrat_date 2020-04-02 Close