摘要
对于红外传感器来说，当距目标很远时，目标在成象平面上仅占一个或几个象素，且接收到的信号很微弱。由于点目标象素少，无形状特征，信噪比低，如何正确检测出目标就是一个值得研究的问题。本文首先对于这种检测问题的主要技术难点进行了讨论，然后分别介绍了目前具有代表性的几种常用方法，例如Ｆａｌｃｏｎｅｒ于１９７７年提出的Ｈｏｕｇｈ变换法，Ｒｅｅｄ等人采用匹配滤波的方法，Ｍｏｈａｎｔｙ等人提出的基于概率估计的最大似然比自适应方法，Ｌａｒｓｏｎ于１９６６年首先提出的动态规划法，Ｂｌｏｓｔｅｉｎ等提出的序贯检测法和近年来所提出的神经网络法，作者对他们进行了评述，最后指出了点目标检测技术今后的发展方向和主要研究内容。
sing infrared sensor,the size of target in the imaging plane of camera is one pixel or several pixels and the signal received is very weak when the target is far away in distance.Without shape feature and low R(SN), it is a difficult problem which defects the target correctly. The main problems existed in this kind of techniques are discussed. Subsequently, a various of currently used method are evaluated, such as: bough transformation method proposed by Falconer etc. in 1977,a matched filter method introduced by Reed etc. an adaptive maximum likelihood ratio based on the estimation of probability presented by Mohanty etc., the dynamic programming method based on target state,vectors first proposed by Larson in 1966,sequential detection method described by Blostein and neural network technique,which is a key point in recent years. Finally the development and main unsolved problems in this area are pointed out.
出处
《数据采集与处理》
CSCD
1994年第4期294-299,共6页
Journal of Data Acquisition and Processing
基金
国家863高技术资助
关键词
目标探测
图象处理
红外图象
检测
target detection
image processing
infrared image
detection techniques
