Electronic Theses and Dissertations





Document Type


Degree Name

Master of Science


Electrical and Computer Engr


Electrical Engineering

Committee Chair

Eddie Jacobs

Committee Member

Srikant K Chari

Committee Member

Aaron L Robinson


This thesis examines the image quality assessment using Structural Similarity Index Metric (SSIM). The performance of Structural Similarity Index Metric was evaluated by comparing Mean Structural Similarity Index (MSSIM) index values with the Probability of Identification (PID) values. The perception experiments were designed for letter images with blur and letter images with blur and noise to obtain the PID values from an ensemble of observers. The other set of images used in this study were tank images for which PID data existed. All the images used in the experiment belong to Gaussian and Exponential filter shapes at various blur levels. All images at a specific blur level and specific filter shape were compared and MSSIM was obtained. MSSIM was compared with blur and PID was compared with blur at various levels for both the filter shapes to observe the correlation between SSIM and human perception. It is noticed from the results that there is no correlation between MSSIM and PID. The image quality differences between SSIM and human perception were obtained in this thesis. From the results it is noticed that SSIM cannot detect the filter shape difference where as humans perceived the difference for letter images with blur in our experiments. The Probability of Identification for Gaussian is lower than the Exponential filter shape which is explained by the edge energies analysis. It is observed that the results of tank images and letter images with blur and noise were similar where humans and MSSIM cannot distinguish between filter shapes.


Data is provided by the student.

Library Comment

Dissertation or thesis originally submitted to the local University of Memphis Electronic Theses & dissertation (ETD) Repository.