[2025] Малый портрет Табуна

from PIL import Image
import os

im = Image.open('D:/tabun/табун 2025.jpg')
width, height = im.size
for j in range(0, width // 48):
    for k in range(0, height // 48):
        im.crop((j*48, k*48, (j+1)*48, (k+1)*48)).save(f'D:/tabun/files/{j+1}-{k+1}.jpg', quality=100)
        if j*k % 100 == 0:
            print(j*k)
input()

import numpy as np
import PIL.Image as Image
from skimage.metrics import peak_signal_noise_ratio as PSNR
from skimage.metrics import structural_similarity as SSIM
import glob
import os
from itertools import islice

def rgb_to_Y(img):
    xform = np.array(
            [[65.738 / 256.0, 129.057 / 256.0, 25.064 / 256.0],
            [- 37.945 / 256.0, - 74.494 / 256.0, 112.439 / 256.0],
            [112.439 / 256.0, - 94.154 / 256.0, - 18.285 / 256.0]])
    img = img.dot(xform.T)[:,:,0:1]+16.0
    return img



if __name__ == "__main__":
    data_dir = "D:/tabun/files"
    images = sorted(glob.glob(data_dir+'/*.jpg'))
    
    target = Image.open("D:/tabun/target.jpg").convert('RGB')
    target_arr = rgb_to_Y(np.array(target).astype(np.float64))

    psnr_ = dict()
    ssim_ = dict()

    for img in images:
        crop = Image.open(img).convert('RGB')
        crop_arr = rgb_to_Y(np.array(hr).astype(np.float64))
        psnr_[img] = PSNR(crop_arr, target_arr, data_range=255.)
        ssim_[img] = SSIM(crop_arr, target_arr, win_size=11, gaussian_weights=True, multichannel=True, channel_axis=-1, K1=0.01, K2=0.03, sigma=1.5, data_range=255)
    
    psnr_sorted = dict(sorted(psnr_.items(), key=lambda item: item[1], reverse=True))
    ssim_sorted = dict(sorted(ssim_.items(), key=lambda item: item[1], reverse=True))
  
    top_psnr = dict(islice(psnr_sorted.items(), 1))
    top_ssim = dict(islice(ssim_sorted.items(), 1))

    print(top_psnr)
    print(top_ssim)

    input()