Airport Runway Detectionon In The Infrared And SAR Remote Sensing Images

In the disposal of remote sensing image, the airport is an important target. It is important to extract and identify the airport runway in military and civil field. The paper studies and realizes the runway detection algorithm for SAR and infrared airport image.Firstly, the paper studies how to remove the noise in infrared image and the speckle in SAR image. Classical filtering methods for infrared image and speckle reduction operators for SAR image are introduced. The experimental results show that the operator based on the threshold wavelet decomposition has a better restraining effect for speckle.Secondly, one-dimensional and two-dimensional OTSU fast threshold segmentation algorithms of airport image are studied. The segmentation algorithm based on Mean Shift is discussed in detail, which can detect the runway quickly and effectively in remote sensing airport image with complex scene. The results show that this method has robustness to some extent and good segmentation effect with the window of suitable radius.Thirdly, the paper introduces the traditional edge-detection operators and improves Canny edge-detection algorithm. An edge-detection algorithm based on confidence level is realized, which has better edge-detection effect. After the extraction of the airport runway, the paper studies two algorithms to extract straight lines. One is phase-based grouping and the other is Hough transform. An improved straight line extraction algorithm based on Hough transform is realized, which can extract the straight-line of the runway edge more effectively.Fourthly, the identification of the runway is studied. Suitable runway model is built based on its characters, which can effectively remove the disturbance of paths and airport-affiliated facilities against the identification of runway. Experimental results show that this model can effectively identify the runway.Finally, a fusion algorithm of SAR and optical airport image based on nonsubsampled contourlet transform is proposed. SAR and optical airport image of the same area are fused according to some rules first, and then edge detection and straight-line extraction for the fused image is carried out. The proposed algorithm is kind of new and practical.