Ncreasing the Speed of Analysis of Images Obtained From Unmanned Aerial Vehicle

Abstract

Carrying out long-term search and rescue missions, intelligence, surveillance and reconnaissance, which are carried out with the help of unmanned aerial vehicles (UAVs) overload the operator and reduce the effectiveness of solving the assigned tasks, up to the point of making their implementation impossible. The paper proposes the concept of increasing the speed of the human-machine system based on information obtained from images transmitted in real time by video cameras installed on a platform moving relative to the UAV, by mean of the intelligent controlsystem of this platform.Thus, the image from the main video camera with a wide viewing angle is divided into rectangular sections (zones), which size is determined by the viewing angle (zoom) necessary for the operator to make the final decision. The software system of the lower level, evaluating the video frame, determines and marks the areas in which the object of interest may be present. Increasing the focal length and focusing (i.e., zooming the image), as well as positioning the main camera is carried out for the area where the probability of the presence of the candidate object is the greatest. In another variant, a separate spot camera with a reduced and fixed field of view is aimed at the yaw and pitch angles, which requires that this camera be directly suspended on a gyro-stabilized platform, where a block of other sensors is installed, including the main video camera.An example of modeling the electric drive of a spot camera by the yaw angle is given and further steps are determined for the development of subsystems of three levels: software (recognition), hardware (scaling and positioning), human-machine (coordination of information flows). To improve the dynamic characteristics in comparison with the basic proportional-integral laws, it is proposed to use quasi-optimal algorithms for regulating electric drives.