干旱区盐渍土全极化SAR极化散射机理与信息识别

In the context of the scarcity of land resources, soil erosion, vegetation degradation, grassland desertification and soil salinization of irrigation canal in arid land of central Asia, to meet the demands for saline soil resources exploration, development and protection in arid lands, this project aiming at saline soil polarimetric scattering mechanism and information identification in arid land environments using PolSAR. Specifically, the advanced cost sensitive learning (CSL) theory is planning to introduce to PolSAR image processing and interpretation to construct the framework of CSL based saline soil information extraction using PolSAR. And, research and solve the corresponding key problems such as saline soil identification feature extraction and construction, cost sensitive learning algorithms, cost strategies and accuracy evaluation. Trough saline soil feature construction, information identified and mappings tasks, propose novel and effective algorithms and processing flow, and evaluate the performances of the proposed algorithms. Finally, provide technical support for saline soil information identification in arid lands via push forward the new PolSAR image processes technique, also promote the applications of CSL in geography, extend the research content and application fields of CSL.

在亚洲中部干旱区土地资源匮乏、土壤侵蚀、植被退化、草地沙化及灌溉渠土壤盐渍化等社会及生态问题突出背景下，围绕干旱区土壤盐渍化调查与盐渍土资源开发利用需求，以干旱区盐渍土全极化SAR极化散射机理分析、盐渍土信息识别与制图为目标，将代价敏感学习理论引入全极化SAR图像处理与信息提取，构建基于代价敏感学习的全极化SAR盐渍土信息识别与信息识别方法体系。研究和解决其中涉及的盐渍土信息识别特征构建、代价敏感学习算法、新型代价敏感学习策略、识别精度评价与算法等关键技术。并围绕盐渍土特征构建、信息识别与制图三个典型任务，提出、建立和形成有效的算法模型和处理流程，通过干旱区盐渍土范围，盐渍化状态和趋势调查、分析、评估与制图等应用实例进行算法性能评价。以期推动全极化SAR图像处理新方法研究，为全极化SAR盐渍土探测与识别应用提供技术支持，促进代价敏感学习的地学应用，丰富代价敏感学习研究内容与应用领域。

在亚洲中部干旱区土地资源匮乏、土壤侵蚀、植被退化、草地沙化及灌溉渠土壤盐渍化等社会及生态问题突出背景下，围绕干旱区土壤盐渍化调查与盐渍土资源开发利用需求，以干旱区盐渍土全极化SAR极化散射机理分析、盐渍土信息识别与制图为目标，将代价敏感学习理论引入全极化SAR图像处理与信息提取，构建基于代价敏感学习的全极化SAR盐渍土信息识别与信息识别方法体系。研究和解决其中涉及的盐渍土信息识别特征构建、代价敏感学习算法、新型代价敏感学习策略、识别精度评价与算法等关键技术。并围绕盐渍土特征构建、信息识别与制图三个典型任务，提出、建立和形成有效的算法模型和处理流程，通过干旱区盐渍土范围，盐渍化状态和趋势调查、分析、评估与制图等应用实例进行算法性能评价。以期推动全极化SAR图像处理新方法研究，为全极化SAR盐渍土探测与识别应用提供技术支持，促进代价敏感学习的地学应用，丰富代价敏感学习研究内容与应用领域。