Gen_Image_method

A toy project to create a generic image processing package.

The project consist of two main parts:

1. Image enhancement via Machine learning, and
2. Error detection and correction methods in image processing.

The aim of this project is to create a set of useful functions and network for more in-depth project related to Astronomical image processing or Atomic force microscpy in the future.

At the moment, few functions on source detection and deconvolution are built.

Quick Examples

Adding bright sparks

import noise_handling as N
from skimage import color, data, restoration
from scipy.signal import convolve2d as conv2

# Load image
Image = color.rgb2gray(data.immunohistochemistry())

# Add sparks to the frame
sparky = N.add_spark(Image, spark_num =20, connect_pix= 400)

# Plot the image+ sparks
N.plot_three_frame(Image,sparky, astro+sparky,'Original Data', 'Sparks', "Composed Image")

Source detection

# Detecting bright sources
bright_spot = N.source_detect(astro+sparky)

# Plot the detection
N.draw_ellipses(astro+sparky, bright_spot,astro)

Creating a noisy convolved image and Deconvolve it

# Generate a Gaussian point-spread-function
psf = N.gaussian_filter(10,sigma=5,muu=0)

# 2D convolution
Image = conv2(Image, psf, 'same')

# Add Noise to Image
Image_noisy = Image.copy()
Image_noisy += N.gen_noise_poisson(255, Image.shape, norm = 255)

# Deconvolution
deconvolved_Image = N.deconvolve(Image_noisy, psf, num_iter=500)

# Plot
N.plot_three_frame(Image,Image_noisy, deconvolved_Image)

# There are much room to be improved on the deconvolution method