Road detection and flight routes determination by using image processing

Road Detection and Flight Routes Determination by Using Image Processing Nowadays, unmanned aerial vehicles (UAV) have an important place. These vehicles are used to provide tracking, monitoring and security. We aim to make control of the vehicle as autonomous without remote control. Thus, an observation which is independent of people and low cost would provide. This study was performed to detection ways to create a route. Thus, it is desirable to provide security by monitoring and tracking. Studies in the literature, generally uses only one of the methods for the detection roads. These methods are as follows: classification, finding the right direction, color and tonal differences. In this study, two of the methods used simultaneously to make the detection roads. Our methods are k-Nearest neighbor classification and Hough transform method. In our algorithm, before we take image from the camera and filter the noise at the image. Then the shadows in the image are detected and removed. Then, the roads are determined with the first method which called nearest neighbor classification method. After this process, the image without shadows is subjected to the second method which called Hough transform methods and roads are detected. Then, the road results which are obtained by the two methods are matched. Determined roads are the final result. After this step, the middle line of the road is detected and the route is identified. Using both methods at the same time with verification increases the success rate. Our work achieves higher performance than the other studies at the determination of the roads. Consequently, the UAV is able to autonomously follow the desired path in accordance with the determined route. Once the destination is reached, it can return to the starting point by following the route which is determined. Thus, flights can be performed without human control and remote control. Road position and route can be determined without GPS data. Keywords: Road Detection, Unmanned Aerial Vehicles, Nearest neighbor, Hough transform, fusion.