Object separation in shadow clutter in video sequences

Identifying moving objects from a video sequence is a fundamental and critical task in many computer-vision applications. Such automatic object detection soft wares have many applications in surveillance, auto navigation, traffic monitoring and robotics. While identifying the objects it is some time essential to identify the objects individually. Individual object detection is very critical in many applications in the areas of surveillance, military, traffic monitoring and medical. The object separation is difficult because of clutter of the objects in the given image frame. Cluttering of the objects is due to overlapping of objects or their shadows making them merge into each other. In videos when multiple moving objects are to be detected its shadow creates clutter in detection phase. Shadows appear in the detection as objects itself if not paid separate attention. In this paper we have demonstrated how shadow affects the separation of the objects in multiple objects detection and method to avoid this object clutter. Here we have separated foreground pixel from background pixel using Gaussian mixture model. These separated pixel forms the mask for next stage of object detection. In cluttered frame the only masked pixels are compared for colour intensity test to detect the cast shadow. This technique reduces the processing time as compared to total pixel testing. Such shadow pixels are removed from the frame and then frame is passed to object detection stage. Shadow removal before object detection stage gives good segmentation of individual object in video having multiple moving objects.