基于梯度比率的SAR图像局部特征提取方法研究
发布时间：2018-12-17 08:12
【摘要】：本文研究了基于像素灰度差值计算的LBP算子和基于梯度比率的LGRP算子等局部二值模式。首先介绍了基本LBP算子和其他几种LBP算子的变形模式,并通过光学图像和实测SAR图像对LBP算子进行性能评估。针对LBP对SAR图像乘性噪声敏感的问题,利用梯度比率计算的LGRP算子,并结合旋转不变LBP的抗旋转性,本文提出了一种改进的SAR图像LGRP特征,获得了对SAR图像的抗噪性和抗旋转性能。实验结果表明,由本文方法提取的SAR图像局部特征具有较好的不变性,可用于姿态角变化下的目标识别与图像纹理切片匹配。
[Abstract]:In this paper, the local binary modes such as LBP operator based on pixel gray difference calculation and LGRP operator based on gradient ratio are studied. Firstly, the basic LBP operator and several other LBP operators are introduced, and the performance of the LBP operator is evaluated by optical images and measured SAR images. Aiming at the problem that LBP is sensitive to multiplicative noise of SAR image, using the LGRP operator calculated by gradient ratio and the anti-rotation property of rotation-invariant LBP, an improved LGRP feature of SAR image is presented in this paper. The anti-noise and anti-rotation performance of SAR images are obtained. The experimental results show that the local features of SAR images extracted by this method have good invariance and can be used for target recognition and image texture slice matching under the change of attitude angle.
【作者单位】： 国防科技大学电子科学与工程学院;
【基金】：国家自然科学基金(61401477)
【分类号】：TN957.52
本文编号：2383940
[Abstract]:In this paper, the local binary modes such as LBP operator based on pixel gray difference calculation and LGRP operator based on gradient ratio are studied. Firstly, the basic LBP operator and several other LBP operators are introduced, and the performance of the LBP operator is evaluated by optical images and measured SAR images. Aiming at the problem that LBP is sensitive to multiplicative noise of SAR image, using the LGRP operator calculated by gradient ratio and the anti-rotation property of rotation-invariant LBP, an improved LGRP feature of SAR image is presented in this paper. The anti-noise and anti-rotation performance of SAR images are obtained. The experimental results show that the local features of SAR images extracted by this method have good invariance and can be used for target recognition and image texture slice matching under the change of attitude angle.
【作者单位】： 国防科技大学电子科学与工程学院;
【基金】：国家自然科学基金(61401477)
【分类号】：TN957.52
【相似文献】
相关期刊论文 前1条
1 项德良;粟毅;赵凌君;唐涛;陆军;;一种基于局部梯度比率特征度量SAR图像相似性的新方法[J];电子学报;2014年01期
，本文编号：2383940
本文链接：https://www.wllwen.com/kejilunwen/xinxigongchenglunwen/2383940.html
