Obstacle detection algorithms for aviation

Wires can be hard to see for pilots and thus present a serious hazard to low-flying aerial vehicles. An airborne vision system having ability to detect wires in time to avoid accident collisions is needed. However, for an automatic vision-based wire detector, extracting wires from a cluttered background is the most important and challenging tasks. In this paper, we propose a novel wire detection algorithm based on real outdoor images for aviation. This detection process combines the morphological properties of the wire and the common line segment detector in image processing. We first generate a line pool using a fast line segment detector to mark all the lines in image. Then wires detector is applied according to our designed properties of wires. The experimental results demonstrate that the approach is effective for automatic wire detection.