仿真条件下贝努利试验的匹配概率估计方法  

Matching probability estimation based on Bernoulli trial on simulation condition

作　　者：王鹏[1] 孙继银[1] 王薇 

机构地区：[1]第二炮兵工程学院,陕西西安710025 [2]二炮驻七零四所代表室,北京100076

出　　处：《激光与红外》2012年第2期209-212,共4页Laser & Infrared

摘　　要：前视红外景象匹配算法性能评价是一个亟待解决的问题。主要研究了匹配概率这一重要算法性能评价指标的估计方法。通过对环境影响因素在红外实时图像上噪声、模糊、放缩和旋转等效应的仿真,进行贝努利匹配试验,估计出单帧图像的匹配概率,并根据单帧图像与部分序列图像匹配概率的关系,计算得到部分序列图像的匹配概率。本文的研究成果为评价前视红外景象匹配算法性能提供了一种有效的新方法。Evaluation of forward looking infrared（FLIR）scene matching algorithm is a problem pressing for solution.This paper focuses on the estimation of matching probability,which is an important indicator of algorithm evaluation.Noise,blur,resizing and rotation of the infrared image,caused by surrounding factors,are simulated.Bernoulli trial is implemented to estimate the matching probability of single image.And then matching probability of partial image sequence is calculated according to the relationship between matching probability of single image and that of the partial image sequence.This paper presents a new method for effective evaluation of FLIR scene matching algorithm.

关 键 词：前视红外 匹配概率 贝努利试验 矩估计 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]