Methods and algorithms for identifying special points in images obtained from an onboard photo video recorder of an unmanned aerial vehicle

The purpose of the research is to develop methods for controlling unmanned aerial vehicles based on the analysis of data coming from a video stream. Methods. An unmanned aerial vehicle can lose contact with the satellite navigation system, so the task of ensuring its orientation using an onboard photo video recorder with onboard data processing becomes relevant. For this purpose, special points on the terrain are used, the identification of which in the picture allows restoring the orientation of the aircraft. To search for special points of the blob type in the picture, a method is proposed for transforming the original image into a criterion image, after threshold processing of which the coordinates of the blobs are obtained. A method has been developed for transforming the original image into a criterion image, which consists in determining correlation images. For each correlation image, a scalar blob identification criterion is determined, which allows determining the coordinates of a special point in pictures obtained from an onboard photo video recorder. Results. To improve the accuracy of determining the coordinates of blobs in pictures, an aggregated blob of three special points was used. An algorithm for two-stage identification of coordinates of an aggregated blob is investigated. At the first stage, the coordinates of special points closest to the coordinates of the vertices of the aggregated blob are determined, and at the second stage, the coordinates of the vertices of the triangle whose center of gravity is closest to the center of gravity of the aggregated blob are determined. The algorithms for searching for special points have shown their efficiency at a high level of interference modeled in the image by means of Gaussian noise, as well as interference associated with the deviation of the aircraft from the specified course. Conclusion. The formation of an aggregated blob with subsequent multi-stage identification allows for increasing the accuracy of determining its coordinates, and also makes it possible to record the deviation from the course of the aircraft in the area of two adjacent images and introduce appropriate corrections into the navigation system.