Research On Airport Runway Detection Method Based On Peak Density Search In PolSAR Images

Synthetic Aperture Radar(SAR)can obtain high-resolution images under low visibility weather conditions,which has the capability of all-day and all-weather.It has been widely used in disaster detection,resource exploration,target detection,ground object classification and other fields.As an aviation infrastructure,the automatic detection of airports is of great strategic significance.However,because polarimetric SAR is easily interfered by radio frequency source in the same frequency band,the image resolution is lower than that of SAR,which has negative impact on target detection.How to make effective use of polarization characteristics of airport runway and improve the accuracy and efficiency of detection is the key to research.The polarization scattering characteristics of airport runway are studied,and two detection methods are given in the thesis.Aiming at the problem that the number of categories needs to be preset in the detection method based on unsupervised classification,a category adaptive method for airport runway detection was proposed.This method can automatically determine the cluster center and cluster number according to the distribution characteristics of ground objects in the images.Firstly,the image is divided into several subcategories by h/q/A decomposition and polarization scattering power,and the polarization similarity matrix between subclasses is established.Then,density peaks clustering and outlier detection are used to determine the cluster center and the number of categories of the images.Subsequently,Wishart fast iterative classifier is used to classify pixel by pixel,and the region of interest(ROI)of the suspected airport runway is extracted.Finally,using the scattering and structural characteristics of runways,the ROIs are identified and the airport runway areas are determined.Aiming at the problem of multiple categories and redundant calculation of classification-based detection methods,an airport runway detection method based on figure segmentation and classification was proposed to improve detection efficiency.Firstly,the pseudo scattering power of the image is defined,and the image is thresholded by using Tsallis entropy and pseudo scattering power to obtain the ROI regions.Then,a classifier combining density peaks clustering and Wishart is used for fine classification of ROIs.Finally,combining with the compactness and contrast characteristics of the airport runway area,the ROIs are identified and the complete airport runway areas are extracted.The experiments were carried out with multiple sets of measured data collected by American UAVSAR,Japanese ALOS-2 and Chinese GF-3 systems.The results show that these two methods can effectively detect all airports with complete structure,clear runway contour and low false and missed alarm rates.The proposed methods have improved detection efficiency and have certain robustness.