CompareImages

CompareImages is a command-line tool that allows you to compare images using various algorithms. You can compare images within a folder or compare them to a black or dark frame.

Prerequisites

• Python 3.6 or higher

Installation

1. Clone the repository:

git clone [email protected]:AdrienLF/ImageComparison.git
cd compare-images

2. Install the required dependencies:

pip install -r requirements.txt

Usage

python CompareImages.py --img-folder <path_to_folder> [--to-black] [--to-dark] [--dark-threshold <threshold>]

Replace <path_to_folder> with the actual path to the folder containing the images you want to compare. Use the optional flags --to-black and/or --to-dark to enable the respective comparison options. The --dark-threshold option allows you to set a custom threshold value for the dark frame comparison (default is 10).

Examples

1. Compare images within a folder:

python CompareImages.py --img-folder img/

Example output:

Comparing credits1.jpg with credits2.jpg:
    ssim = 0.3587869243964623 (Higher means identical)
    mse = 14.930908203125 (Lower means identical)
    rootsift = 21 (Matches between images)
    credits1.jpg borders black = False
    credits1.jpg borders dark = False

0.3587869243964623 21 14.930908203125 False False

2. Compare images to a black frame:

python CompareImages.py --img-folder img/ --to-black

Example output:

Comparing black with credits1.jpg:
            ssim = 0.3587869243964623 (Higher means identical)
            mse = 14.930908203125 (Lower means identical)
            rootsift = 21 (Matches between images)
            black borders black = False
            black borders dark = False
            
0.3587869243964623 21 14.930908203125 False False

3. Compare images to a dark frame with a custom threshold:

python CompareImages.py --img-folder img/ --to-dark --dark-threshold 20

Example output

Comparing dark (threshold = 20) with credits1.jpg:
            ssim = 0.18999967343675722 (Higher means identical)
            mse = 19.7728271484375 (Lower means identical)
            rootsift = 0 (Matches between images)
            dark (threshold = 20) borders black = False
            dark (threshold = 20) borders dark = True
            
0.18999967343675722 0 19.7728271484375 False True

Algorithms

The following algorithms are used for image comparison:

SSIM: Structural Similarity Index. Returns a float between 0 and 1, where 1 means the images are identical.

RootSIFT: Calculates the number of matches between images.

MSE: Mean Squared Error. Returns the difference between images using the Mean Squared Error metric.

Borders: Compares the borders of the images to either black or a dark frame.

License

This project is licensed under the MIT License.