摘要
图像的多尺度分析是对图像进行自底向上的分解,每一层图像均是前一层图像经过某种模板滤波形成的,在大尺度下图像分辨力高,小尺度下分辨力低。为了解决在低对比度下含小目标图像,占绝大部分的均值快变、方差缓变的背景等问题。论文提出采用灰度值插值方法构造图像的多尺度空间,该方法最终可以输出是一个均匀的,能够代表原图像均值的图像。
Multiscale analysis of images is bottom-up decomposition of images,each layer of image is filtered and formed through a certain template from the previous layer of image. The larger the scale is,the higher the image resolution is. In order to solve the problem that the background with quickly changing mean value and slowly changing square deviation predominates in th eimage containing small targets under low contrast ratio,in this study,the multi-scale space of images is constructed by using thegray-value interpolation method. By using the method,a uniform image representing the mean value of the original image can be output at last.
作者
李晓芳
张萍
徐晶
尹福成
LI Xiaofang;ZHANG Ping;XU Jing;YIN Fucheng(Engineering ＆ Technical College,Chengdu University of Technology,Leshan 614000;Neijiang Normal University,Neijiang 641000)
出处
《舰船电子工程》
2018年第8期103-106,共4页
Ship Electronic Engineering
基金
国家高技术研究发展计划(863计划)基金(编号:A08K021)
国家自然科学基金面上项目(编号:61201397
11375055)资助
关键词
多尺度分析
小目标图像
灰度值插值
多尺度逼近图像
multiscale analysis
small targe timage
gray value interpolation
multiscale approximation image
