项目名称: 基于三重判别随机场模型的SAR图像分割研究
项目编号: No.61301284
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 无线电电子学、电信技术
项目作者: 张鹏
作者单位: 西安电子科技大学
项目金额: 26万元
中文摘要: 本项目以非平稳SAR图像分割为研究内容,以SAR图像的非平稳统计特性、统计特征以及SAR地物目标的散射机制为基础,以研究SAR图像的非平稳统计建模为重点,结合贝叶斯信息融合理论、专家场理论以及多分辨率处理理论,研究并建立基于贝叶斯信息融合的三重判别随机场模型的SAR图像分割的新理论和新方法,研究并建立基于三重高维判别随机场模型的SAR图像分割的新理论和新方法,研究并建立基于分层三重判别随机场模型的SAR图像分割的新理论和新方法,突破SAR图像信息的全面有效捕获、SAR数据统计分布模型的自动选取、SAR图像非平稳复杂空域结构的精确统计建模、模型能量最小化的有效统计推导以及特征准确提取等技术难点,提供非平稳SAR图像分割的新理论和新方法,为地球变化研究、地图绘制以及战场侦察和打击评估提供有效的方法和技术支持,具有重大的民用和军事应用价值。
中文关键词: 三重马尔可夫场模型;条件随机场模型;非平稳SAR图像分割;多分辨率分析;贝叶斯信息融合
英文摘要: This project studies the nonstationary SAR image segmentation based on the research of nonstationary statistical property,feature extraction of SAR image and the electromagnetic scattering mechanisms of SAR data. Focusing on the nonstationary statistical modeling of SAR image, this project constructs nonstationary SAR image segmentation theories and methodologies utilizing the triplet discriminative random fields model based on Bayesian fusion theory, establishes nonstationary SAR image segmentation theories and methodologies based on triplet high-order discriminative random fields model and constructs nonstationary SAR image segmentation theories and methodologies based on hierarchical triplet discriminative random fields model. The aim of this project is to capture the SAR image information in a more completed manner effectively, model the statistics of SAR data adaptively in segmentation, extract the statistical feature of SAR image exactly, model the nonstationary complex spatial configuration of SAR image accurately and construct effective Bayesian inference with respect to the proposed statistical model. This project is able to provide the novel theories and methodologies for nonstationary SAR image segmentation and promote SAR applications in many fields such as geological exploration, map updating, detec
英文关键词: triplet Markov fields model;conditional random field model;non-stationary SAR image segmentation;multisclae resolution analysis;Bayesian information fusion