lifting_97

  实现9/7小波正反变换 注1： 采用标准正交方法，对行列采用不同矩阵（和matlab里不同） 注2： 为了保证正交，所有边界处理，全部采用循环处理 注3： 正交性验证，将单位阵带入函数，输出仍是单位阵（matlab不具有此性质） 注4： 此程序是矩阵实现，所以图像水平分量和垂直分量估计被交换位置 注5： 此程序实现的是类小波（wavelet-like）变换，是介于小波包变换与小波变换之间的变换 注6： 此程序每层变换相对原图像矩阵，产生的矩阵都是正交阵，这和小波包一致 注7： 但小波变换每层产生的矩阵，是相对每个待分解子块的正交矩阵，而不是原图像的正交矩阵 注8： 且小波变换产生的正交矩阵维数，随分解层数2分减少 注9: 提升系数可以在MATLAB7.0以上版本，用liftwave( 9.7 )获取，这里直接给出，考虑兼容性 注10：由于MATLAB数组下标从1开始，所以注意奇偶序列的变化 注11：d为对偶上升，即预测；p为原上升，即更新(Note 1: The standard orthogonal method, the ranks of the different matrices (and matlab where different) Note 2: In order to ensure orthogonality, all boundary treatment, all with recycling Note 3: Orthogonal validation, the identity matrix into the function, the output is still the unit matrix (matlab does not have this property) Note 4: This program is a matrix to achieve, so the image horizontal and vertical components is estimated to be swapped Note 5: This class implements the wavelet (wavelet-like) transformation, is between the wavelet packet transform and wavelet transform between Note 6: This procedure transforms each image relative to the original matrix, the resulting matrix are orthogonal matrix, consistent with this and wavelet packet Note 7: But each generation wavelet transform matrix, is to be broken down for each sub-block relative to the orthogonal matrix, rather than the original image of the orthogonal matrix Note 8: Wavelet transf)