红外成像自动目标识别技术研究--计算模型与数据流程

红外自动目标识别是当前智能化图像处理及应用开发的前沿关键技术,其研究进展与计算机视觉的发展水平紧密相关。人类视觉系统是计算机视觉的原始模型,其视觉感知机理的研究将有助于揭示视觉表象的本质,进而为准确描述图像特征信息提供科学而可靠的依据。主要从视觉感知模型、感知功能模块响应特性、视觉对比灵敏度等方面对视觉感知基本原理加以综合分析,并力图利用这些功能卓越的信息处理机制阐明一种具有普适性的视觉计算模型———目标-背景表征模型。在此基础上,将背景区域感知与目标特征分析相结合,提出了自适应信号检测、目标特征识别和运动轨迹跟踪的层次化数据处理流程,从而为红外自动目标识别技术提供一条新的探索途径。

Approach to automatic target recognition in infared imagery--computing model and data flow

Automatic target recognition (ATR) in infrared imagery is the leading key technique of current intelligent image processing and application development. The research progress is consistent with the development level of computer vision. The human vision system is an initial model of computer vision, and the research on perceptive field in human vision could lend its itself to provide us with some important principle about vision phenomena that would offer reliably scientific proof to describe image characteristics.Makes integrative analysis on the perceptive model of vision, the responsive characteristic of perceptive function model and vision contrastive sensitivity. According to this knowledge, a general visual-computing model that describes characters both target and background is established necessarily. Combining perception of background area with analysis of target character, it is presented that a hierarchical data processing flow which includes adaptive signal detection, target character recognition and motive object tracking. It is able to supply new avenue for research on automatic target recognition in infrared image.