Study On Feature Extraction Of Ships Based On ISAR Images

Inverse synthetic aperture radar(ISAR) is a very important high resolution radar. It can work of long range applications in much weathers and times. ISAR plays a very important role in many military and civilian fields. With the improvement of imaging techniques, the sharpness of ISAR image is increasing, ISAR has been widely applied to imaging of the sea surface targets and space targets, ISAR images also contain geometric features and scattering point distribution of targets, which are with great value in target identification, how to obtain information from the target based on ISAR images for target identification classification has become a hot area of research. As a starting point, this dissertation studies on feature extraction of ships based on ISAR images, the main work is summarized as follows:1. The principle of ISAR imaging and signal processing methods of ISAR are introduced, then how to make use of electromagneti simulation software FEKO to do data simulation is studied. Finally experimental verification is done.2. The method of quality assessment of ISAR image is carried on the thorough research,by analyzing the problems of image quality, such like existing speckle noise, stripes interference, ghosting, motion blur, etc. A classification method of image quality is presented in this paper, it can classify images by extracting the characteristics that could reflect the information of image quality. Finally experimental verification is done.3. The extraction methods of length feature and characteristic of attribute scattering center of ships based on ISAR images are mainly studied.(1)For the size of the ship classification applications, we can classify target effectively by extracting its centerline,due to commonly used methods such as Hough transform exists large error in the target centerline extraction, we use PCA transform to extract centerline in this paper, this method uses dimensionality reduction method to find the projection of targets in the main axis.(2)In order to carry out a more detailed classification of ship target, the characteristics of attribute scattering center is carried on, we use orthogonal matching pursuit algorithm to extract attributed scattering center of the target, and the experiment has carried on to verify the validity of the algorithms.