Blur Generator

This tool is for generating blur on images.

There are 3 types of blur modes of motion, lens, or gaussian.

We can use the results on model training or something else.

You can find a simply use case with deep learning in https://github.com/NatLee/simply-blur-detector

Installation

pip install blurgenerator

Check it on Pypi.

Usage

blurgenerator --help

usage: blurgenerator [-h] [--input INPUT] [--input_depth_map INPUT_DEPTH_MAP] [--output OUTPUT] [--type TYPE] [--motion_blur_size MOTION_BLUR_SIZE] [--motion_blur_angle MOTION_BLUR_ANGLE] [--lens_radius LENS_RADIUS] [--lens_components LENS_COMPONENTS]
                     [--lens_exposure_gamma LENS_EXPOSURE_GAMMA] [--gaussian_kernel GAUSSIAN_KERNEL] [--depth_num_layers DEPTH_NUM_LAYERS] [--depth_min_blur DEPTH_MIN_BLUR] [--depth_max_blur DEPTH_MAX_BLUR]
                        Size for motion blur. Default is 100.
  --motion_blur_angle MOTION_BLUR_ANGLE
                        Angle for motion blur. Default is 30.
  --lens_radius LENS_RADIUS
                        Radius for lens blur. Default is 5.0.
  --lens_components LENS_COMPONENTS
                        Components for lens blur. Default is 4.
  --lens_exposure_gamma LENS_EXPOSURE_GAMMA
                        Exposure gamma for lens blur. Default is 2.
  --gaussian_kernel GAUSSIAN_KERNEL
                        Kernel for gaussian. Default is 100.
  --depth_num_layers DEPTH_NUM_LAYERS
                        Layer for depth blur. Default is 3.
  --depth_min_blur DEPTH_MIN_BLUR
                        Min. blur for depth blur. Default is 1.
  --depth_max_blur DEPTH_MAX_BLUR
                        Max. blur for depth blur. Default is 100.

Example and Result

Common use

• Original image

Usage

• Motion blur

blurgenerator --type motion --input ./doc/test.png --output ./doc/motion.png

import cv2
from blurgenerator import motion_blur
img = cv2.imread('test.png')
result = motion_blur(img, size=100, angle=30)
cv2.imwrite('./output.png', result)

• Lens blur

blurgenerator --type lens --input ./doc/test.png --output ./doc/lens.png

import cv2
from blurgenerator import lens_blur
img = cv2.imread('test.png')
result = lens_blur(img, radius=5, components=4, exposure_gamma=2)
cv2.imwrite('./output.png', result)

• Gaussian blur

blurgenerator --type gaussian --input ./doc/test.png --output ./doc/gaussian.png

import cv2
from blurgenerator import gaussian_blur
img = cv2.imread('test.png')
result = gaussian_blur(img, 100)
cv2.imwrite('./output.png', result)

With depth map

Feature from this issue.

• Original image

• Depth map

Usage

• Motion blur with depth map

blurgenerator --input .\doc\depth-test.jpg --type motion --input_depth_map .\doc\depth-map-test.png --depth_num_layers 5 --depth_min_blur 1 --depth_max_blur 50 --output .\doc\depth-motion-output.png

import cv2
from blurgenerator import motion_blur_with_depth_map
img = cv2.imread('test.jpg')
depth_img = cv2.imread('test-depth.png')
result = motion_blur_with_depth_map(
   img,
   depth_map=depth_img,
   angle=30,
   num_layers=10,
   min_blur=1,
   max_blur=50
)
cv2.imwrite('./output.png', result)

• Lens blur with depth map

blurgenerator --input .\doc\depth-test.jpg --type lens --input_depth_map .\doc\depth-map-test.png --depth_num_layers 3 --depth_min_blur 1 --depth_max_blur 50 --output .\doc\depth-lens-output.png

import cv2
from blurgenerator import lens_blur_with_depth_map
img = cv2.imread('test.jpg')
depth_img = cv2.imread('test-depth.png')
result = lens_blur_with_depth_map(
   img,
   depth_map=depth_img,
   components=5,
   exposure_gamma=5,
   num_layers=10,
   min_blur=1,
   max_blur=50
)
cv2.imwrite('./output.png', result)

• Gaussian blur with depth map

blurgenerator --input .\doc\depth-test.jpg --type gaussian --input_depth_map .\doc\depth-map-test.png --depth_num_layers 3 --depth_min_blur 1 --depth_max_blur 50 --output .\doc\depth-gaussian-output.png

import cv2
from blurgenerator import gaussian_blur_with_depth_map
img = cv2.imread('test.jpg')
depth_img = cv2.imread('test-depth.png')
result = gaussian_blur_with_depth_map(
   img,
   depth_map=depth_img,
   sigma=5,
   num_layers=10,
   min_blur=1,
   max_blur=50
)
cv2.imwrite('./output.png', result)

Contributor

Nat Lee

Reference

• Circularly symmetric convolution and lens blur

LICENSE

MIT