The project takes the digitized Prokudin-Gorskii glass plate images and uses image processing techniques to automatically produce a color image with as few visual artifacts as possible.
The objective is to recreate color images from the digitized glass plate photographs Sergey Mikhaliovich Prokudin-Gorskii (1863-1944) took during the last years of the Russian Empire. The primary challenge is to align the three separate color channels—blue, green, and red (BGR)—from these glass plates to produce a single, cohesive RGB image with minimal visual artifacts.
The process involves extracting the color channels, aligning them to form a composite image, and applying image processing techniques to ensure the final image is as clear and accurate as possible. The project includes a basic alignment algorithm using a single-scale search and an advanced version utilizing a coarse-to-fine pyramid approach to efficiently handle larger images (ex. TIFF files).
- split_color_channels(): Splits the image into its BGR components
- crop_image(): Removes unnecessary borders by cropping the images
- get_edges(): Extracts image edges using the Canny algorithm
- SSD(): Calculates the Sum of Squared Differences for alignment scoring
- align() & get_displacement(): Aligns the channels by finding the optimal displacement
- pyramid_align(): Applies a multi-scale approach for efficient alignment of large images
- colorize(): The main function that integrates all the steps to produce and save the final colorized image
Clone the repository to your local machine:
git clone https://github.com/masheharashid/colorize_prokudin_gorskii_collection.git
cd colorize_prokudin_gorskii_collection- Launch the Jupyter Notebook Interface
-
Open the terminal and navigate to the project directory. Then, launch the Jupyter Notebook interface and open the Python notebook file
jupyter notebook colorize_prokudin_gorskii_collection.ipynb
- Set Up Input and Output Image Paths
-
Set the correct paths for results_dir and img_dir in the notebook. Locate the following sections:
-
III. The Main Function: colorize()
# Saves each colorized image in the "colorized_results" folder results_dir = "/path/to/your/colorized_results" # Replace with your actual file path
-
IV. Putting Everything Together
# Path to the "original_data" folder, which contains the input images img_dir = "/path/to/your/original_data" # Replace with your actual file path
-
-
Replace "/path/to/your/colorized_results" and "/path/to/your/original_data" with your local paths
- Follow the instructions in the notebook and run each cell in order
- Upload the Notebook
- Open Google Colab
- Upload the colorize_prokudin_gorskii_collection.ipynb file
- Mount Google Drive (Optional)
-
If your image files are stored in Google Drive, mount the drive to access them:
from google.colab import drive drive.mount('/content/drive')
- Set Up Input and Output Image Paths
- Update the results_dir and img_dir to reflect the paths on Google Drive or within the Colab environment
- Run the Notebook
- Execute the cells in order as instructed in the notebook