ビニング処理をPythonでやってみる

import numpy as np
import matplotlib.pyplot as plt
from PIL import Image, ImageSequence, ImageChops
from scipy.stats import norm
import os

def bin_image(image, bin_size):
    rows, cols = image.shape
    binned_rows = rows // bin_size[0]
    binned_cols = cols // bin_size[1]
    binned_image = np.zeros((binned_rows, binned_cols))
    for i in range(0, binned_rows * bin_size[0], bin_size[0]):
        for j in range(0, binned_cols * bin_size[1], bin_size[1]):
            block = image[i:i + bin_size[0], j:j + bin_size[1]]
            binned_image[i // bin_size[0], j // bin_size[1]] = np.mean(block)
    return binned_image

def plot_histogram_and_get_std(data1, data2, title1, title2):
    plt.hist(data1.ravel(), bins=256, range=(0, 256), color='gray', density=True, alpha=0.5, label=title1)
    plt.hist(data2.ravel(), bins=256, range=(0, 256), color='blue', density=True, alpha=0.5, label=title2)
    plt.title(f'Overlayed Histograms - {title1} & {title2}')
    plt.xlabel('Pixel Value')
    plt.ylabel('Frequency')
    plt.grid(True)
    plt.legend(loc='upper right')
    mu1, std_dev1 = norm.fit(data1.ravel())
    mu2, std_dev2 = norm.fit(data2.ravel())
    return std_dev1, std_dev2

def save_images_as_tif(images, save_path):
    images[0].save(save_path, save_all=True, append_images=images[1:])

# Load the image
image_path = 'image_name_here.jpg'  # Replace with your actual image path
original_image = Image.open(image_path).convert('L')
original_image_array = np.array(original_image)

# Perform binning at multiple bin sizes
bin_sizes = [(2, 2), (4, 4), (8, 8)]
std_dev_results = {}
collected_images = []
collected_histograms = []

for size in bin_sizes:
    binned_image_array = bin_image(original_image_array, size)
    binned_image = Image.fromarray(np.uint8(binned_image_array))
    collected_images.append(binned_image)
    
    plt.figure(figsize=(6, 6))
    std_dev_original, std_dev_binned = plot_histogram_and_get_std(original_image_array, binned_image_array, f'Original - Bin Size: {size}', f'Binned - Bin Size: {size}')
    histogram_image_path = f'histogram_{size[0]}x{size[1]}.png'
    plt.savefig(histogram_image_path)
    plt.close()  # Close the plot after saving
    histogram_image = Image.open(histogram_image_path).convert('L')
    collected_histograms.append(histogram_image)
    
    std_dev_results[size] = (std_dev_original, std_dev_binned)

# Save images and histograms as multi-page TIF files
collected_images_tif_path = os.path.splitext(image_path)[0] + '_collected_images.tif'
collected_histograms_tif_path = os.path.splitext(image_path)[0] + '_collected_histograms.tif'
save_images_as_tif(collected_images, collected_images_tif_path)
save_images_as_tif(collected_histograms, collected_histograms_tif_path)

# Save the standard deviations to a text file with the same name as the input image
output_file_path = os.path.splitext(image_path)[0] + '_multi_bin_std_dev.txt'
with open(output_file_path, 'w') as f:
    for size, (std_o, std_b) in std_dev_results.items():
        f.write(f"Bin Size: {size}\n")
        f.write(f"Standard Deviation of Original Image: {std_o}\n")
        f.write(f"Standard Deviation of Binned Image: {std_b}\n\n")