基于心理认知的高分辨率PolSAR图像智能解译与目标识别方法研究

In the background of ground surface interpretation and of target identification, based on psychological cognition, combining interpretation process of PolSAR image interpretation experts, a high resolution PolSAR image psychological cognition model taking into account of “Information Fusion-Fuzzing Reasoning-Comprehensive Decision Making” is established in the project. Aiming at application requirements, the thinking process, the experience and the utilizing of knowledge and its accumulation of PolSAR image interpretation experts are introduced into image interpretation process, a comprehensive, fast, accurate and automatic high resolution PolSAR image interpretation and target identification algorithm that can be used in sophisticated scene and that aims at application needs is studied. The main contents of the research including: typical surface scattering characteristics and feature extraction of high resolution PolSAR image, psychological cognition process of PolSAR image interpretation experts and established and perfection of high resolution PolSAR image psychological cognition model, PolSAR image interpretation and target identification application based on psychological cognition model. The results of the research are expected to boost the development of high resolution PolSAR image interpretation technology, to fully elaborate the effectiveness of high resolution PolSAR systems in china, and to push forward the application of Chinese high resolution PolSAR monitoring in both national economy and national defense.

本课题以地物解译与目标识别为应用背景，以心理认知理论为基础，结合PolSAR图象专业解译人员的认知过程，构建包括“信息融合-模糊推理-综合决策”的高分辨率PolSAR图象的心理认知模型。针对应用需求，课题将PolSAR图象专业解译人员的思维过程、经验与知识运用及积累过程引入到图象解译处理，研究适用于复杂场景，面向需求并具有一定普适性的高分辨率PolSAR图象全面、快速、准确、智能化解译与目标识别算法。主要研究内容包括：高分辨率PolSAR图象典型地物散射特性研究及有效特征提取，PolSAR图象专业判图人员心理认知过程及高分辨率PolSAR图象心理认知模型构建和完善，基于心理认知模型的PolSAR图象解译与目标识别应用。研究成果以期促进高分辨率PolSAR图象解译技术的发展，充分发挥我国高分辨率PolSAR系统的效益，推动我国高分辨率PolSAR对地观测系统在国民经济和国防建设领域的应用。

本项目针对高分辨率PolSAR图像智能化解译的需求，为实现高水平、智能化的典型地物目标检测识别、力求将人类信息处理与专业判图员图象解译能力迁移至PolSAR图像自动处理上，项目从高分辨率PolSAR图象典型地物散射特性出发，研究了PolSAR图象有效信息提取与表征方法，并结合人类心理认知理论及思维过程，研究了PolSAR图象专业判图解译人员的心理认知过程，构建了高分辨率PolSAR图象心理认知模型，并基于心理认知模型实现了高分辨率PolSAR图象解译与目标识别。.本项目以人类认知建模与专业判图解译人员的认知过程为理论基础，结合PolSAR图象地物的散射特点，以高分辨率PolSAR图象智能化解译为应用需求，具体实现了高分辨率PolSAR图象典型目标检测与识别，解决了其中人类认知过程的数学描述以及知识的形式化表征、认知过程的学习机制与专业判图员PolSAR图像认知模型构建等关键科学问题，构建了PolSAR心理认知模型，推动了人类心理认知在PolSAR图象解译中的发展，能够为实现人类认知水平的PolSAR图象解译提供重要的参考价值。.本项目取得的技术成果经过8种星载与机载PolSAR图象的19景实测数据的实验验证，本项目中提出的方法与成果具有高分辨率PolSAR图象的不同场景下典型地物的认知能力。本项目其他成果包含了24篇高水平期刊论文以及9篇国际会议论文、申请了7项专利、培养了12名博士与27名硕士研究生。