Digital images are frequently contaminated by impulse noise(IN)during acquisition and transmission.The removal of this noise from images is essential for their further processing.In this paper,a two-staged nonlinear f...Digital images are frequently contaminated by impulse noise(IN)during acquisition and transmission.The removal of this noise from images is essential for their further processing.In this paper,a two-staged nonlinear filtering algorithm is proposed for removing random-valued impulse noise(RVIN)from digital images.Noisy pixels are identified and corrected in two cascaded stages.The statistics of two subsets of nearest neighbors are employed as the criterion for detecting noisy pixels in the first stage,while directional differences are adopted as the detector criterion in the second stage.The respective adaptive median values are taken as the replacement values for noisy pixels in each stage.The performance of the proposed method was compared with that of several existing methods.The experimental results show that the performance of the suggested algorithm is superior to those of the compared methods in terms of noise removal,edge preservation,and processing time.展开更多
遥感影像在采集过程中,地面覆盖种类数量庞大且采集影像清晰度低、分辨率较差,关键像素特征之间的阈值衡量标准模糊,导致信息提取难度增大,从而降低信息利用率。为此,提出了基于像素紧密程度的多源遥感影像信息提取方法。利用Contourle...遥感影像在采集过程中,地面覆盖种类数量庞大且采集影像清晰度低、分辨率较差,关键像素特征之间的阈值衡量标准模糊,导致信息提取难度增大,从而降低信息利用率。为此,提出了基于像素紧密程度的多源遥感影像信息提取方法。利用Contourlet变换,实现遥感影像空间域、变换域的多角度增强,优化遥感影像整体清晰度。利用简单线性迭代聚类(Simple Linear Iterative Clustering,SLIC)超像素算法计算像素聚类中心与邻近像素紧密程度,摆脱固定阈值影响。引入灰度共生矩阵(Gray-level Co-occurrenceMatrix,GLCM),提取主体特征信息;构建相关向量机分类模型,结合拉普拉斯二次逼近回归算法将提取问题转化为噪声回归问题,并对其展开求解,进而实现遥感影像的信息提取。实验结果表明:所提方法对遥感信息主体的分类与真实遥感信息主体分类基本一致,在信息提取过程中的错提取率和漏提取率低,总体提取精度保持在99%以上,且对道路信息提取清晰度高,表明该方法提高了遥感信息的解译水平。展开更多
基金supported by the Opening Project of Key Laboratory of Astronomical Optics & Technology, Nanjing Institute of Astronomical Optics & Technology, Chinese Academy of Sciences (No. CAS-KLAOTKF201308)partly by the special funding for Young Researcher of Nanjing Institute of Astronomical Optics & Technology,Chinese Academy of Sciences(Y-12)
文摘Digital images are frequently contaminated by impulse noise(IN)during acquisition and transmission.The removal of this noise from images is essential for their further processing.In this paper,a two-staged nonlinear filtering algorithm is proposed for removing random-valued impulse noise(RVIN)from digital images.Noisy pixels are identified and corrected in two cascaded stages.The statistics of two subsets of nearest neighbors are employed as the criterion for detecting noisy pixels in the first stage,while directional differences are adopted as the detector criterion in the second stage.The respective adaptive median values are taken as the replacement values for noisy pixels in each stage.The performance of the proposed method was compared with that of several existing methods.The experimental results show that the performance of the suggested algorithm is superior to those of the compared methods in terms of noise removal,edge preservation,and processing time.
文摘遥感影像在采集过程中,地面覆盖种类数量庞大且采集影像清晰度低、分辨率较差,关键像素特征之间的阈值衡量标准模糊,导致信息提取难度增大,从而降低信息利用率。为此,提出了基于像素紧密程度的多源遥感影像信息提取方法。利用Contourlet变换,实现遥感影像空间域、变换域的多角度增强,优化遥感影像整体清晰度。利用简单线性迭代聚类(Simple Linear Iterative Clustering,SLIC)超像素算法计算像素聚类中心与邻近像素紧密程度,摆脱固定阈值影响。引入灰度共生矩阵(Gray-level Co-occurrenceMatrix,GLCM),提取主体特征信息;构建相关向量机分类模型,结合拉普拉斯二次逼近回归算法将提取问题转化为噪声回归问题,并对其展开求解,进而实现遥感影像的信息提取。实验结果表明:所提方法对遥感信息主体的分类与真实遥感信息主体分类基本一致,在信息提取过程中的错提取率和漏提取率低,总体提取精度保持在99%以上,且对道路信息提取清晰度高,表明该方法提高了遥感信息的解译水平。
