如何批量扫描图片文件夹，基于图像均值识别暗图

I see you already have the core logic to detect dark images using the mean pixel value—great start! Let's expand that into a full batch processing script that scans your entire folder, identifies dark images, and lets you handle them (like logging their paths or moving them to a separate directory).

Full Batch Processing Code #

from skimage import io, img_as_float
import numpy as np
import os
from glob import glob

# Configuration - adjust these paths and thresholds to your needs
IMAGE_FOLDER = "C:/Data/Testing"
DARK_IMAGE_THRESHOLD = 0.1
# Optional: Create a folder to move dark images to (avoids accidental deletion)
DARK_IMAGE_DEST = os.path.join(IMAGE_FOLDER, "dark_images")
os.makedirs(DARK_IMAGE_DEST, exist_ok=True)

# Get all image files in the folder (supports common formats; add more if needed)
image_paths = glob(os.path.join(IMAGE_FOLDER, "*.jpg")) + glob(os.path.join(IMAGE_FOLDER, "*.png")) + glob(os.path.join(IMAGE_FOLDER, "*.jpeg"))

# Iterate through each image
for img_path in image_paths:
    try:
        # Load and preprocess the image (same as your single-image code)
        image = io.imread(img_path)
        image = img_as_float(image)
        mean_value = np.mean(image)
        
        # Classify the image
        if mean_value < DARK_IMAGE_THRESHOLD:
            print(f"Dark image detected: {img_path} (mean: {mean_value:.4f})")
            # Optional: Move the dark image to the dedicated folder
            os.rename(img_path, os.path.join(DARK_IMAGE_DEST, os.path.basename(img_path)))
        else:
            print(f"Normal image: {img_path} (mean: {mean_value:.4f})")
    except Exception as e:
        print(f"Error processing {img_path}: {str(e)}")

Key Details Explained #

• Folder Traversal: We use glob.glob() to target specific image formats (JPG, PNG, JPEG) instead of scanning all files—this avoids trying to process non-image files which would cause errors.
• Error Handling: The try-except block catches issues like corrupted images or unsupported file types, so the script won't crash halfway through.
• Organized Storage: The optional dark_images folder lets you segregate dark images instead of just logging them, which is safer than deleting them directly.
• Configurable: You can easily adjust IMAGE_FOLDER, DARK_IMAGE_THRESHOLD, or add more image formats to the glob calls to match your setup.

Quick Notes #

• If you don't want to move the images and just need a list of dark ones, comment out the os.rename() line and instead append the paths to a text file for later reference.
• For very large folders, consider adding a progress tracker (like using the tqdm library) to see how many images have been processed.

内容的提问来源于stack exchange，提问作者Zephyr India