Carried baggage detection and recognition in video surveillance with foreground segmentation
Year: 2014
Type: dissertation
Abstract: Security cameras installed in public spaces or in private organizations continuously\nrecord video data with the aim of detecting and preventing crime. For that reason,\nvideo content analysis applications, either for real time (i.e. analytic) or post-event\n(i.e. forensic) analysis, have gained high interest in recent years. In this thesis,\nthe primary focus is on two key aspects of video analysis, reliable moving object\nsegmentation and carried object detection & identification.\nA novel moving object segmentation scheme by background subtraction is presented\nin this thesis. The scheme relies on background modelling which is based\non multi-directional gradient and phase congruency. As a post processing step,\nthe detected foreground contours are refined by classifying the edge segments as\neither belonging to the foreground or background. Further contour completion\ntechnique by anisotropic diffusion is first introduced in this area. The proposed\nmethod targets cast shadow removal, gradual illumination change invariance, and\nclosed contour extraction.\nA state of the art carried object detection method is employed as a benchmark\nalgorithm. This method includes silhouette analysis by comparing human temporal\ntemplates with unencumbered human models. The implementation aspects of\nthe algorithm are improved by automatically estimating the viewing direction of\nthe pedestrian and are extended by a carried luggage identification module. As\nthe temporal template is a frequency template and the information that it provides\nis not sufficient, a colour temporal template is introduced. The standard\nsteps followed by the state of the art algorithm are approached from a different\nextended (by colour information) perspective, resulting in more accurate carried\nobject segmentation.\nThe experiments conducted in this research show that the proposed closed\nforeground segmentation technique attains all the aforementioned goals. The incremental\nimprovements applied to the state of the art carried object detection\nalgorithm revealed the full potential of the scheme. The experiments demonstrate\nthe ability of the proposed carried object detection algorithm to supersede the\nstate of the art method.
Author: Giounona Tzanidou
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W750017806