An Efficient System on Morphological Operations to Detect Vehicles

With the development of technology, automated vehicle detection in aerial surveillance keeps on improving with congestion avoidance, toll collection, military, police and security and surveillance system. It has become an important field to study. Different technologies have been proposed and implemented for detection of vehicles in an aerial view, which covers a large area. The paper proposes a new idea on vehicle detection based on the morphological operation. Firstly image is converted into HSV format. HSV format image is split into three separate channels. Opening and closing transformation is performed on each channel followed by top hat and bot hat transformation. In the process, bright and dark objects are segmented by top hat and bot hat transformation respectively. Segmented results of both transformations which extracted bright and dark object are added so that none of the vehicles are left behind. Static noise is removed from the added image. 8Way connectivity neighboring is performed to group the connected pixel and is labeled separately to each group. Finally, post-processing operations are carried out to determine the vehicle. The results of the paper demonstrate very effective approach in detection of a vehicle on different aerial videos taken under different camera angles and heights.