Paper Title
Building A Quad Robot for Image Processing and its Controller to Capture Images from the Environment
Abstract
Due to its special design compared to other helicopters, Chaharmalkh or the same four-propeller helicopter has a
more stable flight, which makes it suitable for surveillance and aerial photography, especially in UAV research (unmanned
aerial vehicle for gathering information and Taking pictures ) has been used a lot in recent years. Today, four-propeller
helicopters are used in various conditions, and in many cases, they are used as a robot with direct user intervention, as a
control element in the control loop. Using a human force to control these robots, in addition to the overhead cost that it
entails, is not desirable in many circumstances. In times when the robot is used to monitor city traffic, which requires
guiding the robot behind buildings, and if there are blind spots, there will be a possibility of losing control of the robot. The
purpose of this article is to build a quadcopter with a controller to process the image on the video sent by the quadcopter.
Image processing is done by CNN algorithm or convolutional neural network with deep learning and by opencv library. The
results implemented by the opencv library show that the designed system performs facial recognition with an accuracy of
over 90%. In the system database for image processing, we have used two data samples including about 6,000 human body
images and 5,000 non-human body images collected for the purpose of this article.
Keywords - Opencv, Image Processing, Deep Learning, Quadcopter, Controller, Convolutional Neural Network