针对条带模式合成孔径雷达回波缺失数据,提出了一种利用压缩感知恢复缺失数据并成像的方法。将条带数据分块为多个子孔径数据,对子孔径利用压缩感知恢复缺失数据并拼接得到条带数据,缩短了整个数据的恢复时间,推导了压缩感知处理的基矩...针对条带模式合成孔径雷达回波缺失数据,提出了一种利用压缩感知恢复缺失数据并成像的方法。将条带数据分块为多个子孔径数据,对子孔径利用压缩感知恢复缺失数据并拼接得到条带数据,缩短了整个数据的恢复时间,推导了压缩感知处理的基矩阵和测量矩阵。运用最大似然估计的特征向量方法(eigenvector method for maximum likelihood estimation,EMMLE)实现了子孔径缺失数据的自聚焦,满足了压缩感知对图像的稀疏要求。利用压缩感知恢复完整的相位误差信号,解决了子孔径补偿相位误差数据的拼接问题。最后通过对恢复的雷达回波数据成像并自聚焦校正了距离徙动,得到了聚焦良好的完整图像,提高了缺失数据的成像质量。展开更多
In recent years, the popular multifractal detrended fluctuation analysis （MF-DFA） is extended to two-dimensional （2D） version, which has been applied in some field of image processing. In this paper, based on the ...In recent years, the popular multifractal detrended fluctuation analysis （MF-DFA） is extended to two-dimensional （2D） version, which has been applied in some field of image processing. In this paper, based on the 2D MF-DFA, a novel multifractal estimation method for images, which we called the local multifractal detrended fluctuation analysis （LMF-DFA）, is proposed to recognize and distinguish 20 types of tea breeds. A set of new multifractal descriptors, namely the local multifractal fluctuation exponents is defined to portray the local scaling properties of a surface. After collecting 10 tea leaves for each breed and photographing them to standard images, the LMF-DFA method is used to extract characteristic parameters for the images. Our analysis finds that there are significant differences among the different tea breeds＇ characteristic parameters by analysis of variance. Both the proposed LMF-DFA exponents and another classic parameter, namely the exponent based on capacity measure method have been used as features to distinguish the 20 tea breeds. The comparison results illustrate that the LMF-DFA estimation can differentiate the tea breeds more effectively and provide more satisfactory accuracy.展开更多
文摘针对条带模式合成孔径雷达回波缺失数据,提出了一种利用压缩感知恢复缺失数据并成像的方法。将条带数据分块为多个子孔径数据,对子孔径利用压缩感知恢复缺失数据并拼接得到条带数据,缩短了整个数据的恢复时间,推导了压缩感知处理的基矩阵和测量矩阵。运用最大似然估计的特征向量方法(eigenvector method for maximum likelihood estimation,EMMLE)实现了子孔径缺失数据的自聚焦,满足了压缩感知对图像的稀疏要求。利用压缩感知恢复完整的相位误差信号,解决了子孔径补偿相位误差数据的拼接问题。最后通过对恢复的雷达回波数据成像并自聚焦校正了距离徙动,得到了聚焦良好的完整图像,提高了缺失数据的成像质量。
文摘In recent years, the popular multifractal detrended fluctuation analysis （MF-DFA） is extended to two-dimensional （2D） version, which has been applied in some field of image processing. In this paper, based on the 2D MF-DFA, a novel multifractal estimation method for images, which we called the local multifractal detrended fluctuation analysis （LMF-DFA）, is proposed to recognize and distinguish 20 types of tea breeds. A set of new multifractal descriptors, namely the local multifractal fluctuation exponents is defined to portray the local scaling properties of a surface. After collecting 10 tea leaves for each breed and photographing them to standard images, the LMF-DFA method is used to extract characteristic parameters for the images. Our analysis finds that there are significant differences among the different tea breeds＇ characteristic parameters by analysis of variance. Both the proposed LMF-DFA exponents and another classic parameter, namely the exponent based on capacity measure method have been used as features to distinguish the 20 tea breeds. The comparison results illustrate that the LMF-DFA estimation can differentiate the tea breeds more effectively and provide more satisfactory accuracy.
