Spatially-Varying-Blur-Detection-python

python implementation of the paper "Spatially-Varying Blur Detection Based on Multiscale Fused and Sorted Transform Coefficients of Gradient Magnitudes" - cvpr 2017

brief Algorithm overview

Uses discrete cosine transform coefficients at multiple scales and uses max pooling on the high frequency coefficients to get the sharp areas in an image.

Quickstart

This library performs Spatially Varying Blur Detection which is can be used in many applications such as Depth of field estimation, Depth from Focus estimation, Blur Magnification, Deblurring etc.

Installation

To install, run: pip install blur-detector

Usage:

import blur_detector
import cv2
if __name__ == '__main__':
	img = cv2.imread('image_name', 0)
	blur_map = blur_detector.detectBlur(img, downsampling_factor=4, num_scales=4, scale_start=2, num_iterations_RF_filter=3, show_progress=True)

	cv2.imshow('ori_img', img)
	cv2.imshow('blur_map', blur_map)
	cv2.waitKey(0)

As easy as that!!