Aerospace Instrument-Making Annotation << Back ALGORITHM FOR FINDING THE HORIZON LINE IN IMAGES TAKEN FROM AN UNMANNED AERIAL VEHICLE CAMERA Yu.G. Emelyanova The work solves the problem of detecting the horizon line in images of the area obtained from the camera of an unmanned aerial vehicle. The purpose of the study is to improve the quality of detection of natural fires by determining the horizon line and transforming the original image on this basis. A method adapted to reduced visibility in a natural fire is proposed. The method is based on calculating a local contrast equivalent to an information value of image pixels. The size of the scanning window is set individually for each image by applying a multiple linear regression model. Features characterizing the features of the survey conditions were selected as model factors. An algorithm has been developed for merging the points of the horizon line into a single set. Keywords: horizon line, unmanned aerial vehicle, natural fire, weather conditions, multiple linear regression model, local contrast of the image. DOI: 10.25791/aviakosmos.1.2023.1318 Pp. 40-53.