摘要
在分析红外图象和可视图象差异的基础上 ,研究了一种适用于可视图象序列的运动小目标检测算法 .受目标影响的象素点具有较弱的时间相关性 ,即相对普通象素点有较大的时间方差 ,根据此特征 ,用管道更新的方法产生新方差图象序列 ,再用管道更新的方法累积 ,增加信噪比 ,然后用基于统计均值的自适应门限方法分割出目标 ,最后将原图象的梯度倒数作为象素点强度的加权 ,有效地抑制灰度突变边界的传感器噪音 。
It describes a novel algorithm for detecting impurity one kind of small target inliquid by analyzing sequences of gray scale images,which is able to detect impurity with full intelligent.It is based on the basic theories of tracking and detection of small target in image sequences and is divided into three main stages:image pre processing,potential targets extraction and target tracking.Each of them is explained in detail in this paper.In the second stage,a new method of extacting features of small targets in visible image sequences is proposed,which is based on the idea that a pixel that is affected by a point target will have greater temporal intensity variance.By calculating the variance summation sequence,the S/N sensitivity of the algorithm is greatly improved.The technique responds to targets of either polarity.Tests of this system on available database have yielded good results.
出处
《光子学报》
EI
CAS
CSCD
北大核心
2002年第6期743-747,共5页
Acta Photonica Sinica
