Detecting and tracking drones using imaging sensors with background subtraction and neural networks

In passive radar technology, a combination of image processing and information science is used which is capable of rapidly detecting the moving target in the picture and specify its location to other ends such as tracking and detecting aims. In these systems, in the first phase, the extraction and the defragmentation of an area of the video image is done. Then the processing technique is repeated by shaking the window throughout the page. The problem becomes more complicated when the subject being extracted is moving and is being shared frequently and in several frames. Besides, due to the relocation of the image according to the photographer, image properties could change to in different times, and this might trouble the correspondent between images in different times and make it hard to accurately extract the path and speed of the movement. Tracking objects on the move in the Continue the frames, because it is involved in many problems in the real world, is of high importance.In this paper, we have been able to detect and track quadcopters using a combination of background subtraction and neural networks. We first coded in MATLAB software then practically processed it using imaging sensors and achieved excellent results.

کلیدواژه ها:

deep learning ، neural network ، object detection ، object tracking