In the forthcoming 5G systems, new technologies, such as amorphous networks and non-orthogonal filter bank multicarriers(FBMC), provide an effective way to accommodate high-rate transmissions. Meanwhile, the prototype...In the forthcoming 5G systems, new technologies, such as amorphous networks and non-orthogonal filter bank multicarriers(FBMC), provide an effective way to accommodate high-rate transmissions. Meanwhile, the prototype filter affects the adjacent channel interference, and therefore is important for FBMC systems. Besides, once the amorphous network is taken into account, the requirement for interference controlmust be much stricter. Accordingly, this paper focuses on the design of prototype filter with better ability of interference controlling, where we exploit the nonlinear phase FIR filter(NLPFF) instead of traditional linear phase FIR filter(LPFF) to achieve more optimization spaces under a small sacrifice of linear phase. In ourdesigns, both the amplitude and phase responsesare handled independently to approach the stopband performance enhancements, in while the nearly perfect reconstruction(NPR) conditionsare relaxed by pre-specified thresholds. Computer simulations confirm the effectiveness of the NLPFF designs, and demonstrate the advantages of the proposed NLPFF in FBMC applications.展开更多
针对不同模态图像存在严重的几何差异和非线性辐射差异(NID)等问题导致的匹配难题,提出了一种结合多尺度特征与局部采样描述的多模态图像配准方法。首先,引入非线性扩散方程构建非线性尺度空间,然后结合相位一致性与ORB(oriented FAST a...针对不同模态图像存在严重的几何差异和非线性辐射差异(NID)等问题导致的匹配难题,提出了一种结合多尺度特征与局部采样描述的多模态图像配准方法。首先,引入非线性扩散方程构建非线性尺度空间,然后结合相位一致性与ORB(oriented FAST and rotated BRIEF)算法获得多尺度的稳定特征点。接着提出了一种旋转不变的双重高斯采样描述符,可以在NID存在的情况下鲁棒地跨越[0°,360°)的旋转差异。最后,引入了一种图像恢复策略。通过初次匹配获得最优的几何变换模型,修正图像间存在的几何差异,再进行二次匹配提升匹配精度。在遥感、医学和计算机视觉等领域的多模态数据集上进行实验,当图像存在尺度、旋转等几何差异时,所提方法均方根误差平均可达到1.5个像素以内,正确匹配率达到98%以上。表明该方法能够克服图像间非线性辐射差异的影响,可以实现高精度配准。展开更多
基金supported by National Natural Science Foundation of China under Grants No.61471322
文摘In the forthcoming 5G systems, new technologies, such as amorphous networks and non-orthogonal filter bank multicarriers(FBMC), provide an effective way to accommodate high-rate transmissions. Meanwhile, the prototype filter affects the adjacent channel interference, and therefore is important for FBMC systems. Besides, once the amorphous network is taken into account, the requirement for interference controlmust be much stricter. Accordingly, this paper focuses on the design of prototype filter with better ability of interference controlling, where we exploit the nonlinear phase FIR filter(NLPFF) instead of traditional linear phase FIR filter(LPFF) to achieve more optimization spaces under a small sacrifice of linear phase. In ourdesigns, both the amplitude and phase responsesare handled independently to approach the stopband performance enhancements, in while the nearly perfect reconstruction(NPR) conditionsare relaxed by pre-specified thresholds. Computer simulations confirm the effectiveness of the NLPFF designs, and demonstrate the advantages of the proposed NLPFF in FBMC applications.
文摘针对不同模态图像存在严重的几何差异和非线性辐射差异(NID)等问题导致的匹配难题,提出了一种结合多尺度特征与局部采样描述的多模态图像配准方法。首先,引入非线性扩散方程构建非线性尺度空间,然后结合相位一致性与ORB(oriented FAST and rotated BRIEF)算法获得多尺度的稳定特征点。接着提出了一种旋转不变的双重高斯采样描述符,可以在NID存在的情况下鲁棒地跨越[0°,360°)的旋转差异。最后,引入了一种图像恢复策略。通过初次匹配获得最优的几何变换模型,修正图像间存在的几何差异,再进行二次匹配提升匹配精度。在遥感、医学和计算机视觉等领域的多模态数据集上进行实验,当图像存在尺度、旋转等几何差异时,所提方法均方根误差平均可达到1.5个像素以内,正确匹配率达到98%以上。表明该方法能够克服图像间非线性辐射差异的影响,可以实现高精度配准。
