Recognition and tracking of moving objects in the images captured by UAV intelligently in earth observation operations

摘要

Today, drone is used for different purposes. Among these, security through detecting and tracking moving vehicles that have violated traffic rules, developing photo coverage from vast and high-risk areas, helping out deprived areas, and so on can be mentioned. In this paper, a new method is presented for detecting and tracking moving objects from moving platforms. The process itself consists of two parts: (1) identifying the path range that the object moves, which helps eliminate excess margins and increase the speed of operations and (2) identifying moving objects inside the path range and tracking them with a new method, which helps increase the accuracy in this regard. The approach presented to the selected video is applied to a frame as large as (1280x720) and has yielded very satisfactory results. Thus, the ability to identify the given moving objects in this method is 66% and 75% in very crowded scenes and in very quiet scenes, respectively. However, the power of the conventional methods in this regard is about 21% and 50% in the scenes with very busy backgrounds and in the scenes with very quiet backgrounds, respectively.