首页> 外国专利> INPAINTING BASED SIGNAL GAP RECONSTRUCTION WITH SPARSE REPRESENTATION

INPAINTING BASED SIGNAL GAP RECONSTRUCTION WITH SPARSE REPRESENTATION

机译:基于稀疏表示的基于信号间隙重建

摘要

For reconstructing a signal (x), such as an audio signal, within a gap where the signal is invalid, a sparse inpainting method is used in relation to dictionary signals which have been modified based on the signal in a neighborhood around the gap. A set of initial dictionary signals (g) is provided and deformed into modified dictionary signals (ĝ) which are defined within the signal domain surrounding the gap, by calculating the modified dictionary signals as invertible linear combinations of the initial dictionary signals, according to an invertible deformation matrix. Analysis coefficients for the input signal are calculated with respect to the initial dictionary signals, to obtain an analysis coefficient vector, and a deformation matrix is determined such that it optimizes a vector norm of a transformed coefficient vector which results from transforming the vector of the analysis coefficients through the matrix. Using the deformation matrix, and/or modified dictionary signals calculated from the deformation matrix, in the sparse inpainting method a remodeled output signal (x) is constructed.
机译:为了重构信号无效的间隙中的信号(x),例如音频信号,在信号无效的间隙内,稀疏的初始化方法与基于间隙周围的邻域中的邻域中的信号进行修改的字典信号。根据一个可逆变形矩阵。关于初始词典信号计算输入信号的分析系数,以获得分析系数向量,并且确定变形矩阵,使得它优化变换系数矢量的矢量范数,这导致分析的载体变换通过矩阵系数。使用从变形矩阵计算的变形矩阵和/或修改的字典信号,在稀疏的初始化方法中,构造了改造的输出信号(X)。

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号