An enhancement of people counting System using edge flow vector Segmentation and fuzzy logic

Abstract

Visual surveillance has been a very active research topic in the last newlinefew years due to its growing importance in matters of security Our goal is to newlinetrack and count multiple moving people observed from static surveillance newlinecameras Methods of background subtraction are widely used in these kinds of newlineconditions But the people count often fails due to noise foreground shadow newlinescene occlusion blob merge and blob split newlineIn this thesis the methods for background estimation background newlinesubtraction segmentation and tracking have been investigated Four methods newlineare proposed to deal with the related problems The automatic estimation of newlinebackground is a challenging job due to variations in the video and the newlinecomplexity of the background In this context the extraction of the newlinebackground from images and videos is an important research problem and newlinereceives growing attention newlineThe first method uses the background model to estimate the newlinebackground image from a video A novel background estimation algorithm newlinebased on an improved mode algorithm is used to obtain the regions of a static newlinebackground The goal of employing these approaches is to obtain a clean newlinestatic background reference image and then applying it for background newlinesubtraction For high frame rate sequences the adjacent frame subtraction newlinemethod is used since the change of motion between the consecuting frames is newlinevery small This method eliminates the stationary background leaving only newlinethe desired motion regions newline newline