SAR Building Detection Based On Morphology And DS Evidence Theory

Synthetic Aperture Radar(SAR)is a kind of active microwave remote sensing sensor.Compared to traditional optical sensor,SAR has the characteristics of working all-day and all-weather,which can get richer scatter information.With the rapid development of China's economy and urbanization,resource allocation,land use,environmental protection,urban rational planning and other issues are highly valued nowadays,which make the demand for information on urban geography changes grow rapidly.And one of the most important needs is the extraction of building information in the city.For these reasons,a SAR image filtering algorithm based on gamma distribution and morphological alternating sequence filtering and a SAR Image building object verification algorithm based on D-S Evidence Theory and Visual Attention Model are proposed in this paper,which can be effectively applied to SAR image building object detection.The specific improvement ideas are as follows:Based on the traditional morphological alternating sequence filtering algorithm and the basic morphological operation,the shape of the structural elements in the alternating sequence is improved.And combined with the characteristics that the area of the SAR image conforms to the gamma distribution,the image is reconstructed based on the gamma distribution after the each iteration of the alternate sequence filter,so that more details information of the image can be preserved.Based on the SLIC super pixel segmentation algorithm,the gray scale morphological reconstruction top-hat operation will be used for larger background area filtering,the Kmeans clustering algorithm is used to classify the obtained super pixels into two categories so as to segment the SAR image.Based on the D-S evidence theory,the characteristics of the building object on the SAR image and the characteristics of the building in the visual attention model as multiple evidence,the confidence of the building hypothesis is calculated by integrating the various features to complete the validation of the building's assumptions.Experiments show that the verification method proposed in this paper can effectively accomplish the judgment of thebuilding hypothesis,can effectively detect building objects in SAR images and has the advantage of reducing the false alarm rate.