ImageCompressor

import numpy as np
import pandas as pd
from sklearn.decomposition import TruncatedSVD
import matplotlib.pyplot as plt
import matplotlib.image as mpimg



class ImageCompressor:
    def __init__(self, n_components):
        self.n_components = n_components
        self.svd = TruncatedSVD(n_components=self.n_components)

    def fit_transform(self, img):
        img_red = img[:, :, 0]
        img_green = img[:, :, 1]
        img_blue = img[:, :, 2]

        components_red = self.svd.fit_transform(img_red)
        compressed_red = np.dot(components_red, self.svd.components_)

        components_green = self.svd.fit_transform(img_green)
        compressed_green = np.dot(components_green, self.svd.components_)

        components_blue = self.svd.fit_transform(img_blue)
        compressed_blue = np.dot(components_blue, self.svd.components_)

        compressed_red = compressed_red.astype(int)
        compressed_blue = compressed_blue.astype(int)
        compressed_green = compressed_green.astype(int)

        compressed_array = np.stack((compressed_red, compressed_green, compressed_blue), axis=2)
        compressed_array = np.clip(compressed_array, 0, 255)

        return compressed_array



img = mpimg.imread('IMG_2844.jpg')
df = img[:, :, 0]
img0 = plt.imshow(df, cmap='grey')
plt.title('Оригінальне зображення')
plt.show()

svd = TruncatedSVD(n_components=df.shape[1])
svd.fit(df)

eigenvalues = svd.singular_values_

plt.bar(range(1, 21), eigenvalues[:20])
plt.xlabel('Номер власного значення')
plt.ylabel('Власне значення')
plt.title('Гістограма перших 20 власних значень')
plt.show()

explained_variance_ratio = svd.explained_variance_ratio_
cumulative_explained_variance = np.cumsum(explained_variance_ratio)

plt.plot(range(1, len(cumulative_explained_variance) + 1), 100 - cumulative_explained_variance * 100, marker='o')
plt.xlabel('Кількість використаних компонент')
plt.ylabel('Втрата інформації, %')
plt.show()

n_components_to_plot = [1, 3, 5, 10, 15, 20]

fig, axs = plt.subplots(2, 3, figsize=(15, 10))
axs = axs.ravel()

for i, n in enumerate(n_components_to_plot):
    svd = TruncatedSVD(n_components=n)
    components = svd.fit_transform(df)
    reconstructed_df = pd.DataFrame(np.dot(components, svd.components_))
    axs[i].imshow(reconstructed_df, cmap="gray")
    axs[i].set_title(f'Кількість компонент = {n}')

plt.show()






plt.title('Оригінальне зображення')
plt.imshow(img)
plt.show()

n_components_to_plot = [1, 3, 5, 10, 15, 20]

fig, axs = plt.subplots(2, 3, figsize=(15, 10))
axs = axs.ravel()

for i, n in enumerate(n_components_to_plot):
    compressor = ImageCompressor(n_components=n)
    reconstructed_img = compressor.fit_transform(img)
    axs[i].imshow(reconstructed_img)
    axs[i].set_title(f'Кількість компонент = {n}')

plt.show()