基于主题模型的SAR图像分类-电路与系统专业论文.docxVIP

ABSTRACT III ABSTRACT With the development of SAR(Synthetic Aperture Radar) technology, the data collection capability of SAR is growing rapidly, it is difficult for visual interpretation to meet the rapid growth of SAR data. Automated or semi-automated imagery exploitation for SAR images through computer and pattern recognition technology can tremendously improve efficiency of data processing, and has a promising future and application value in not only military but also civil field. SAR image classification. it is more useful and has a supreme future. More and more researchers are engaged in this area. SAR image not only contains the target information, but also contains the scene information. The traditional image classification method based on low-level features can not make full use of the scene information, So in recent years researchers are looking for more effective SAR image processing methods. The topic model which has got a great development in the natural image processing in recent years, could provide us with a new tool. In topic model, the completed scene image can be represented with visual vocabularies which express one or more topics, and these topics reflect the scene of the image. In this thesis, we start with SAR image classification problem, use the LDA topic model to set up SAR image scene topic model. First, we study the generation model of the visual vocabulary and topic model, then analyze the influence of different low-level features on SAR image scene topic model, and at last propose the SAR image classification method based on topic model. Set up SAR image scene topic model. This paper make in-depth discussion for SAR image theme model affect by the low layer characteristics, and analysis influence that different characteristics of the visual vocabulary of the image classification. The focus of this paper is mainly to study the characteristics, so we analyze visual vocabulary that are generated by different features. The used low-level fe