• Python3
• opencv-python
• PyWavelets

Install these using the following command:

$ pip install -r requirements.txt

To run the python script with the sample images uploaded to this repo.

python blur_wavelet.py -i images/blur

The paper defines two parameters in order to configure the algorithm. The first is threshold. It is used to select if a pixel of Haar transform image is considered as Edge Point. Default value is 35. If you select a smaller threshold, it is more likely an image to be classified as blur.

The default threshold is 35. You can define it by adding the parameter in the call:

python blur_wavelet.py -i images/noblur --threshold 25

In the paper it is called MinZero. If Per is smaller than MinZero the image is classified as blur. The default value is 0.001 . In order to configure the MinZero threshold, run the script with the flag -d

python blur_wavelet.py -i images/noblur -d 0.005

In order to save the output as .JSON, run the script with the flag -s SAVE_PATH.json .

python blur_wavelet.py -i images/blur -s output.json

The sample images have been taken from this image dataset.

This algorithm is based entirely on this paper