Estimation of the Structural Similarity Index (SSIM) for Image Quality Assessment

Supplementary material to Statistical estimation of the structural similarity index for image quality assessment by Felipe Osorio, Ronny Vallejos, Wilson Barraza, Silvia Ojeda and Marcos A. Landi (Signal, Image and Video Processing 187, 108885. DOI: 10.1007/s11760-021-02051-9)

Code written by: Felipe Osorio and Wilson Barraza

Correspondence author: Felipe Osorio, Email: felipe.osorios@usm.cl

Code tested on:

• R Under development (2021-05-12 r80288), running MX Linux 19.4 (64 bits)
• R under development (2018-02-21 r74285), running Linux Mint 18.3 (64 bits)
• R version 3.3.0, running OS X 10.13.4 (64 bits)

Instructions:

To create the Dynamically Loaded (DL) library, using the console prompt move to /code/src directory and enter:

R CMD SHLIB -o SSIM.so *.c

Next, copy SSIM.so file to the working directory (in our case to /simulation), and execute the commands in simulation.R and CDF.R files.

CONTENTS:

• case_study/ : SAR imagery from Copeland, Kansas; Corpus Christi, Texas and the Mississippi river delta in the state of Louisiana, USA and an image taken from Ullum's dam in Argentina (in .PNG format).
• case_study/SSIM_img.R : R function for fitting an heteroscedastic nonlinear regression model using several filters
• case_study/case_study.R : R function to perform the SSIM parameter estimation on SAR imagery described at Section 3.2 from manuscript.
• code/src/ : C sources for parameter estimation and computation of the structural similarity index for images (SSIM).
• code/SSIM.R : R function for computation of SSIM.
• code/SSIM_FIT.R : R function for parameter estimation of SSIM.
• simulation/gamma.simul.R: R function to perform the contamination with multiplicative noise using a Gamma distribution, for different decreasing levels of contamination, that is, 1, 2, 4, 8, 16, and 32 looks.
• simulation/simulation.R: R commands to perform the simulation study described at Section 3.1 from manuscript.
• simulation/CDF.R: R commands to computation of empirical cumulative distribution of the SSIM under H0 and H1 for several looks.
• data/ : reference images (Lena, Baboon and texmos2) used in numerical experiments (available in .RDA and .PNG formats).
• README.md : this file.