We propose an edgesegmentbased statistical\r\nbackground modelling algorithm to detect the moving\r\nedges for the detection of moving objects using a static\r\ncamera. Traditional pixel intensitybased background\r\nmodelling algorithms face difficulties in dynamic\r\nenvironments since they cannot handle sudden changes\r\nin illumination. They also bring out ghosts when a\r\nsudden change occurs in the scene. To cope with this\r\nissue,intensityandnoiserobustedgebasedfeatureshave\r\nemerged. However, existing edgepixelbased methods\r\nsuffer from scattered moving edge pixels since they\r\ncannot utilize the shape.Moreover, traditional segment\r\nbasedmethodscannothandleedgeshapevariationsand\r\nmiss moving edges when they come close to the\r\nbackground edges. Unlike traditional approaches, our\r\nproposed method builds the background model from\r\nordinary training frames that may contain moving\r\nobjects.Furthermore,itdoesnotleaveanyghostsbehind.\r\nMoreover, our method uses an automatic threshold for\r\nevery background edge distribution for matching. This\r\nmakes our approach robust to illumination change,\r\ncameramovementandbackgroundmotion.Experiments\r\nshowthatourmethodoutperformsothersandcandetect\r\nmoving edges efficiently despite the above mentioned\r\ndifficulties.
Loading....