Python implementation of the Fast Wavelet Transform (FWT) on 1D, 2D, and 3D(soon) input signals/data. The common wavelets like Haar, and Daubechies is available, along with 60+ wavelets.

The code is according to the software development process, so hopefully its user-friendly or dev-friendly.

The simple Wavelet Transform is given by the formula

The fundamental idea of wavelet transforms is that the transformation should allow only changes in time extension, but not shape. This is affected by choosing suitable basis functions that allow for this. Changes in the time extension are expected to conform to the corresponding analysis frequency of the basis function.

Dimension implemented (1D, 2D) Just call waveDec for wavelet decomposition for any dim, length array And waveRec for wavelet reconstruction for any dim, length array

Update: Use it with any length of data. (1D & 2D)

Check the examples/ for some examples on the usage. Refer the html docs/

1. Install using pip

pip install git+https://github.com/AP-Atul/wavelets

2. Clone the repo and run setup

git clone https://github.com/AP-Atul/wavelets.git
python setup.py install

1. Wavelet decomposition and reconstruction

from wavelet import FastWaveletTransform

WAVELET_NAME = "db4"
t = FastWaveletTransform(WAVELET_NAME)

# original data
data = [1, 1, 1, 1, 1, 1, 1, 1]

# decomposition --> reconstruction
coefficients = t.waveDec(data)
data = t.waveRec(coefficients)

2. Simple discrete transforms

from wavelet import WaveletTransform, getExponent

transform = WaveletTransform(waveletName="db2")
data = [1, 2, 3, 4, 5, 6, 7, 9]

# dwt with max level
coefficients = transform.dwt(data, level=getExponent(len(data)))

# inverse dwt with max level
data = transform.idwt(coefficients, level=len(coefficients))

(I'll try to provide some examples for this)

1. Audio de-noising by cleaning the noise signal from the coefficients
2. Data cleaning in the sense of Data Mining
3. Data compression
4. Digital Communications
5. Image Processing
6. etc.

The performance can be improved. Help to make it even better by contributing