C++ Implementation to Compare Pairs of Image Quality using RMSE, SSIM, and PSNR

Compare the quality between two images using RMSE, SSIM, and PSNR. The values of the PSNR can be predicted from the SSIM and vice-versa. The PSNR and the SSIM mainly differ on their degree of sensitivity to image degradations.

Usage

The executable file is under x64/Release/Compare.exe

Compare image_file_name_1 image_file_name_2 [--mask] [--block_size] 2

• The optional mask parameter will neglect the total white or black pixels in the image1.
• The optional block_size parameter determines the window size of SSIM.
• The smaller block_size is, the slower it runs.

Compilation

If the compilation fails, please fix the following environment variables:

• OPENCV_INC: Directory to OpenCV include folder.
• OPENCV_LIB: Directory to OpenCV libraries.
• PATH: Add the executable DLLs of OpenCV and GLUT into an arbitrary directory of PATH.

Details

RMSE: Root Mean Squared Error

The square root of the mean squared error. More sensitive than SSIM and PSNR in values. Wiki: https://en.wikipedia.org/wiki/Root-mean-square_deviation

SSIM: Structural Similarity Index Measure

The SSIM index is calculated on various windows of an image. Wiki: https://en.wikipedia.org/wiki/Structural_similarity

PSNR: Peak signal-to-noise ratio

PSNR is most often defined via 10 multiplies the log10 of the normalized mean squared error (MSE). Wiki: https://en.wikipedia.org/wiki/Peak_signal-to-noise_ratio

Free software!

Contact: Ruofei Du (http://www.duruofei.com)