An analysis of spectral similarity measures

Code for the paper An analysis of spectral similarity measures. In the paper the most used measures for the assessment of spectral similarity are analyzed. Please read the paper for details.

Dependencies

• Python 3.9
• numpy
• scikit-image

In order to successfully run the code, install the packages listed in requirements.txt as follows:

pip install -r requirements.txt

Citation

If you use our code, please consider cite the following:

• Mirko Agarla, Simone Bianco, Luigi Celona, Raimondo Schettini, and Mikhail Tchobanou. An analysis of spectral similarity measures. In Color and Imaging Conference, volume 2021, Society for Imaging Science and Technology, volume 2021, number 6, pp. 300-305, 2021.

  @inproceedings{agarla2021spectralmeasures,
   author = {Agarla, Mirko and Bianco, Simone and Celona, Luigi and Schettini, Raimondo and Tchobanou, Mikhail},
   year = {2021},
   title = {An analysis of spectral similarity measures},
   organization = {Society for Imaging Science and Technology},
   booktitle = {Color and Imaging Conference},
   volume = {2021},
   number = {6},
   doi = {https://doi.org/10.2352/issn.2169-2629.2021.29.300},
   pages = {300--305},
  }
  

Available measures

Mean error measures

• Mean Square Error (MSE)
• Root Mean Square Error (RMSE)
• Mean Relative Absolute Error (MRAE)
• Back-Projection MRAE (BPMRAE)
• Peak Signal-to-Noise Ratio (PSNR)

Similarity measures

• Goodness-of-Fit Coefficient (GFC)
• Mean Structural SImilarity Measure (MSSIM)

Angular measures

• Average Per-Pixel Spectral Angle (APPSA)
• Mean Angular Error (MAngE)
• Spectral Angle Mapper (SAM)

Colorimetric error measures

• Delta E*
• Euclidean distance in ProLab color space

Other measures

• Mean Spectral Information Divergence (MSID)

Acknowledgement

This research was supported by Huawei Technologies Co. Ltd. Russia.