Improved-Image-Compression-with-KMeans/compress_images.py at main · pmephin/Improved-Image-Compression-with-KMeans · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
## RUN THIS FOR A COMPARATIVE STUDY

from compress_utils import *
import pandas as pd

image=io.imread('data/test_image.jpeg')  ##Image path

K_vals=[32,64,128]  #k1 vals

res_reg={'k_vals':K_vals,'PSNR':[],'SSIM':[]}
res_eig={'k_vals':K_vals,'PSNR':[],'SSIM':[]}
res_var={'k_vals':K_vals,'PSNR':[],'SSIM':[]}

for k in K_vals:

    im=compress_reg(image,k)
    io.imsave(f'data/results/reg_k{k}.jpeg',im)
    res_reg['PSNR'].append(peak_signal_noise_ratio(image,im))
    res_reg['SSIM'].append(structural_similarity(image,im,win_size=3))

    im=compress_eig(image,k1=k)
    io.imsave(f'data/results/eig_k{k}.jpeg',im)
    res_eig['PSNR'].append(peak_signal_noise_ratio(image,im))
    res_eig['SSIM'].append(structural_similarity(image,im,win_size=3))

    im=compress_var(image,k1=k)
    io.imsave(f'data/results/var_k{k}.jpeg',im)
    res_var['PSNR'].append(peak_signal_noise_ratio(image,im))
    res_var['SSIM'].append(structural_similarity(image,im,win_size=3))

pd.DataFrame(res_reg).to_csv('data/results/res_reg.csv',index=False)
pd.DataFrame(res_eig).to_csv('data/results/res_eig.csv',index=False)
pd.DataFrame(res_var).to_csv('data/results/res_var.csv',index=False)