Minimal C++ implementation of 1D and 2D Wavelet transform. The interfaces resemble MATLAB wavedec and wavedec2. This is implemented in the Wavelet class, which is able to compute multi-level wavelet decomposition and reconstruction of 1D and 2D data.
An auxiliary Matrixclass is also provided to represent 2D data, whereas std::vector are used to represent 1D data. An additional vector.h header overloads standard mathematical binary and unary operators for std::vector.
Example usage for 1D and 2D data is provided in main.cpp. Run make to compile and execute the program, called sw. The program expects three arguments as follows
./sw num_rows num_cols num_levels
which will be used to initialise a random matrix having size num_rowsxnum_cols, and a Wavelet transform with Haar transform filers to compute a 1D and 2D decomposition with num_levels.
The matrix will be used as input for the 2D decomposition, whereas the first column of the matrix (having length num_rows) will be used as 1D data. Both 1D and 2D data will be reconstructed from their decomposition, and the reconstruction error (l2-norm) will be finally printed to standard output. Note that the seed for the random value generation is fixed, thus the results will be
./sw 64 64 4
Error (l2-norm) of 1D reconstruction: 3.30226e-15
Error (l2-norm) of 2D reconstruction: 5.0788e-14
MIT Licence. Copyright (c) 2018 Matteo Maggioni