| Adalat Pashayev, Ulviyya Goyushova, Sadiq Malikov On the road identification method from images |
|---|
| Abstract. The research work is devoted to the method of autonomous navigation of unmanned aerial vehicles based on video images. The main essence of the navigation method based on video images is to detect important infrastructure objects from the image, determine the features and compare them with the features located in the database. In the study, winding roads without intersections are selected as important infrastructure. Roads are determined from pre-taken images in the base, and only some features about these roads are selected and stored. The UAV takes an image of the earth's surface during its movement. Detects roads from the image, matching features related to these roads are found. The characteristics placed in the database are compared with the characteristics found at the current time. In the study, the roads in the image are identified as straight lines. For this, two different versions of the same method have been proposed. Roads from the image are defined as straight lines. As a result, the winding road is expressed as intersecting straight lines. The powers of the found paths - the number of pixels - points included in the roads are found. The ratio of these numbers to each other is considered as a feature that expresses the image. Software for solving the problem is developed and tested on images. |
| Keywords: UAV images, autonomous navigation, edge detection, line detection, GPS, feature |
 Download PDF |
| DOI: https://doi.org/10.54381/itta2024.38 |