Real time method for airport runway detection in aerial images

In this paper, we propose a novel method that combines improved chain codes based edge tracking (ICCBET) with Hough transform (HT) to successfully detect the airport runway in real time. ICCBET primarily removes the short and curving lines to reduce the pixels and confirms the approximate orientation to shorten the angle range HT processes in later. Furthermore, through using pyramid during HT stage, the computation cost is reduced considerably to satisfy real time performance. Finally, we fix the memory by designing a chain list array for the image to avoid overflow arise from building chain list with memory dynamic allocation. Experiments on various images show that our method satisfies the real time need with about 23.5 multiples computation reduction contrast to the original Hough transform (OHT) and overcomes the blur to localize the lines with high accuracy contrast to the original chain codes based edge tracking (OCCBET).