Robust Multi Object Tracking Removing Shadow Using Fore Ground Detection

Abstract

Multiple objects tracking is a combination of 2 problems one is fragmentation and the other grouping. To overcome these problems blob measurement of bounding boxes and area of each is calculated. This helps in object fragmenting or grouping. For this purpose foreground extraction and back ground subtraction were followed by blob area identification. Even the area ratio helps in counting the vehicles in particular frame. Using this graph and a generic object model based on spatial connectedness and coherent motion, we label the tracked blobs as whole objects, fragments of objects or groups of interacting objects. The outputs of our algorithm are entire tracks of objects, which may include corresponding tracks from groups during interactions. This is mostly useful in real time applications, such as video surveillance, for experimental a traffic video under surveillance was considered and results for this video is also observed.