项目名称: 仿蝇类视觉系统信息获取与处理机制的高分辨率遥感小目标检测研究
项目编号: No.61271386
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 无线电电子学、电信技术
项目作者: 黄凤辰
作者单位: 河海大学
项目金额: 74万元
中文摘要: 针对目前高分辨率遥感影像小目标检测存在的问题和检测方法的局限性,受蝇类视觉系统信息获取和处理机制的启发,与以往研究策略和角度不同,试图建立仿蝇视觉感知信息的表征模型,作为高分辨率遥感小目标检测器及其算法设计的理论依据,转化蝇类视觉系统在复杂多变自然场景下快速、准确捕捉小目标的优势。本课题通过对蝇视觉系统结构、机理及与工程化模拟蝇系统内在机制的比较研究,构建遥感影像目标检测仿蝇视觉信息处理的系统模式。基于此模式,模拟蝇视觉信息获取、压缩、融合的机制,建立基于"弹药筒"的仿蝇视觉感知信息的表征模型;模拟蝇视觉"池细胞"功能和神经元非线性自适应特性,提出基于分流抑制机理的特征层信息重构算法;提出和设计小目标检测器及其算法。目前,高分辨率遥感小目标检测和识别在许多领域有着广泛和迫切的应用需求,开展本项研究有助于推进遥感影像自动解译技术的进步。
中文关键词: 遥感信息处理;目标检测;仿蝇视觉;仿生系统建模;智能信息处理
英文摘要: In view of the problems and the limitation in small target detection of high resolution remote sensing images, this project try to build the representation model of perceptual information inspired by fly visual system which completely different from the existing strategies. The representationmodel can be the theoretical basis of designing the detector of small target detection of high resolution remote sensing image and takes the advantages of fly visual system which can quickly and accurately detect the small targets in complex scenes in nature. Through the comparative study about the structure, mechanism and engineering simulation of the internal mechanism of fly visual system, this project constructs a system model of information processing of target detection in remote sensing. Based on this model, this project simulates the mechanism of information acquisition, compression and fusion by fly visual system, builds the representation model of perceptual sensory information as of the fly visual system based on "cartridge"; This project proposes the method of information reconstruction in feature level based on split-flow and suppression by simulating the function of "pool cells" and the neuron characteristics of nonlinearity and self-adaptation. This project proposes and designs the method and detector for smal
英文关键词: Remote sensing information processing;Target detection;Fly’s visual system;Bionic system modeling;Intelligent information processing