摘要
针对降维解构分析的多尺度图像处理容易造成高维信息的破碎与损失这一问题,建立了一种基于图像几何矩的非降维连续尺度解构模式,并在此模式下构建了10组非降维解构模型。模型以单通道图像为应用对象,以几何矩为算子,以不同尺度窗口下矩运算产生的特征值为基础对原始信息进行非降维解构。10组解构结果呈现均衡模式和增长模式两种形态。代入多尺度分割算法验证证明,基于图像几何矩的非降维连续解构分析可以提升多尺度分割精度,且增长模式下的解构信息对于图像分割更有利。
Dimensionality-reduction analyses are effective to deconstruct image completely,but also destructive for high-dimension information of image.Therefore,this study proposes a 2-dimension multiscale image deconstruction mode to solve this problem which is based on geometric moment.In this mode,10 groups of 2-dimension multiscale deconstruction models are built.Models use single channel image as target,use different geometric moments as operators,use different scales of windows as operand,and use moment calculation to get eigenvalues.Based on eigenvalues,multiscale models are built.According to statistics analysis,10 models are in two patterns,which are equilibrium pattern and growth pattern.Result shows that 2-dimension multiscale image deconstruction analysis could improve the accuracy of multiscale image segmentation,and growth pattern models are more effective than equilibrium pattern models.
作者
曹地
曹建农
朱倩
CAO Di;CAO Jiannong;ZHU Qian(School of Earth Science and Resources,Chang’an University,Xi’an710054,China;Key Laboratory of Degraded and Unused Land Consolidation Engineering,the Ministry of Land and Resources,Xi’an710054,China;College of Geology Engineering and Geomatics,Chang’an University,Xi’an710054,China)
出处
《遥感信息》
CSCD
北大核心
2019年第1期44-50,共7页
Remote Sensing Information
基金
国家自然科学基金(41571346)
国土资源部退化及未利用土地整治工程重点实验室开放基金(SXDJ2017-10
2016KCT-23)
中央高校基本科研业务费专项资金(310826171014)
中央高校基本科研业务费专项资金(300102268647)
关键词
多尺度分析
几何矩
塔式模型
图像非降维解构
图像分割
multiscale analysis
geometric moment
tower model
2D image deconstruction
image segmentation
