结合BIC准则和ECM算法的可变类SAR影像分割
发布时间：2018-12-22 08:23
【摘要】：为实现合成孔径雷达(SAR)影像分割中类别数的自动确定,提出一种基于贝叶斯信息准则(BIC)的可变类SAR影像分割算法.该算法以Gamma分布建模SAR影像同质区域内部像素光谱测度的统计分布特性;结合BIC准则构建整幅SAR影像似然函数模型;并在此模型中引入类别数补偿项,继而提高BIC测度对影像分割结果的描述精度.采用期望条件最大化(ECM)算法估计其模型参数;通过遍历所有可能类别数,取最小BIC值对应的类别数作为最佳类别数.采用提出的算法分割模拟和真实SAR影像,模拟SAR影像分割结果的定性和定量分析表明,基于BIC准则的ECM算法可以实现类别数的自动确定,并可得到最优分割结果.通过对真实SAR影像分割结果的定性评价,进而证明了可变类SAR影像分割算法的准确性和可行性.
[Abstract]:In order to automatically determine the number of categories in (SAR) image segmentation of synthetic Aperture Radar (SAR), a variable class SAR image segmentation algorithm based on Bayesian Information Criterion (BIC) is proposed. The algorithm uses Gamma distribution to model the statistical distribution of the pixel spectral measure in the homogeneous region of SAR image, and combines the BIC criterion to construct the likelihood function model of the whole SAR image. In this model, the category number compensation term is introduced to improve the accuracy of the BIC measure in describing the image segmentation results. The expected condition maximization (ECM) algorithm is used to estimate the model parameters, and the number of classes corresponding to the minimum BIC value is taken as the best category number by traversing all possible categories. The qualitative and quantitative analysis of the segmentation results of simulated SAR images using the proposed algorithm shows that the ECM algorithm based on the BIC criterion can automatically determine the number of categories and obtain the optimal segmentation results. Through the qualitative evaluation of the real SAR image segmentation results, the accuracy and feasibility of the variable class SAR image segmentation algorithm are proved.
【作者单位】： 辽宁工程技术大学测绘与地理科学学院遥感科学与应用研究所;
【基金】：国家自然科学基金青年基金项目(41301479) 辽宁省自然科学基金项目(2015020090)
【分类号】：TN957.52
本文编号：2389563
[Abstract]:In order to automatically determine the number of categories in (SAR) image segmentation of synthetic Aperture Radar (SAR), a variable class SAR image segmentation algorithm based on Bayesian Information Criterion (BIC) is proposed. The algorithm uses Gamma distribution to model the statistical distribution of the pixel spectral measure in the homogeneous region of SAR image, and combines the BIC criterion to construct the likelihood function model of the whole SAR image. In this model, the category number compensation term is introduced to improve the accuracy of the BIC measure in describing the image segmentation results. The expected condition maximization (ECM) algorithm is used to estimate the model parameters, and the number of classes corresponding to the minimum BIC value is taken as the best category number by traversing all possible categories. The qualitative and quantitative analysis of the segmentation results of simulated SAR images using the proposed algorithm shows that the ECM algorithm based on the BIC criterion can automatically determine the number of categories and obtain the optimal segmentation results. Through the qualitative evaluation of the real SAR image segmentation results, the accuracy and feasibility of the variable class SAR image segmentation algorithm are proved.
【作者单位】： 辽宁工程技术大学测绘与地理科学学院遥感科学与应用研究所;
【基金】：国家自然科学基金青年基金项目(41301479) 辽宁省自然科学基金项目(2015020090)
【分类号】：TN957.52
【相似文献】
相关期刊论文 前5条
1 王智顺;毕厚杰;朱梅英;;ECM标准的规程性能分析[J];南京邮电学院学报;1993年03期
2 尹杰;杨魁;段毅;;面向对象的极化干涉SAR影像分割算法研究[J];测绘通报;2013年06期
3 ;对抗反舰巡航导弹的ECM方案[J];航天电子对抗;1994年01期
4 欧阳毅,朱桐华,解希顺,郑虎鸣;ECM使用寿命及FEP电阻率的估算[J];电声技术;2005年05期
5 徐才宏;通用ECM评估方法[J];舰船电子对抗;1995年05期
，本文编号：2389563
本文链接：https://www.wllwen.com/kejilunwen/xinxigongchenglunwen/2389563.html
