Visual-based object tracking with four-rotor aerial vehicle

The need for unmanned aerial vehicles in both the military and civil sectors has increased significantly in recent years. Today, unmanned aerial vehicles are used in missions, such as search and rescue, damage detection, and reconnaissance, as well as in military areas such as intelligence and surveillance. The four-rotor unmanned aerial vehicle of unmanned aerial vehicles with a wide variety of models comes to the forefront with its high mobility. In the thesis, an image-based object tracking of a four-rotor unmanned aerial vehicle was carried out. First of all, a mathematical model of a four-rotor unmanned aerial vehicle was created. Parrot AR.Drone 2.0 Power Edition of Parrot is used for object tracking. In the object tracking process, the color, shape and coordinate information of the object to be tracked are used by using the images taken from the front camera of the device. In the Python programming language, the desired movements of the unmanned aerial vehicle have been achieved with code fragments written using AT (Attention) commands. By using OpenCV which is an important platform in image processing, color feature and shape feature are examined and processed separately in detecting moving object. In order to detect moving objects from images obtained from fixed cameras, color and shape features were used together to create a program that performs point and core based object tracking in the Python programming language. The generated program was connected to the four-rotor unmanned aerial vehicle via wifi, using the images obtained from the front camera of the device, and visual-based object tracking was performed. By connecting with AR.Drone 2.0 through the Autoflight program, the route line that the device was watching while moving was examined. In the EZBuilder program, images taken from the camcorder on AR.Drone 2.0 in the RGB (Red-Green-Blue)color space, a red moving object, in a HLS (Hue- Luminance-Saturation) color space, a yellow moving object and in the YCbCr (Luminance-Chrominance blue-Chrominance red) color space, yellow and green rectangular objects introduction and follow-up procedures were carried out. Key Words: Quadrotor Unmanned Air Vehicle, Image Processing, Object Tracking