Extended distance transform approach for robust vehicle detection

Image and video processing is indispensable for modern traffic surveillance applications. At this, the reliable detection of vehicles is an essential and also challenging task depending on versatile environment parameters. Many research works have been investigated in accurate pattern matching so far. Nevertheless, dealing with noise and variations in pattern shapes are still improvable key problems. This paper presents a novel approach enabling robust detection of vehicles in static frames. The proposed algorithm extends the classical distance transform approach and provides vehicle-specific clutter depression methodologies. In this regard, experimental results are given on diverse traffic scenarios.