Events Details

Speaker:
Mohsen Jenadeleh (Associated Fellow / Computer and Information Science)

Abstract:

Perceptual quality assessment has been a long-standing problem that attracts attention from both academia and industry. The just-noticeable-difference (JND) methodology has been proposed to measure the human subjective quality of experience in recent years. In this talk, I will provide a brief review of JND-based visual quality measurement and its applications. I will also discuss the biases introduced by current methodologies for subjective JND assessment and introduce a methodology to collectively assess the JND and address the bias problem. During this talk, I will explore ways to optimize the JND subjective test methodology in the context of video compression, aiming to estimate user satisfaction with a given video quality at a lower cost and with higher accuracy using both simulations and human studies. Experimental results will be presented to demonstrate the performance of the proposed methods.

Biography:

Mohsen Jenadeleh received his Ph.D. degree from the Department of Computer and Information Science at the University of Konstanz, Germany, in 2019. He is currently a postdoctoral researcher at the University of Konstanz, where he is working on his research project titled "JND-based perceptual video quality analysis and modeling," funded by the German Research Foundation (DFG). His research interests include image processing, visual perception, machine learning, deep learning, and crowdsourcing. Mohsen is also an associated fellow at Zukunftskolleg,and serves as an expert in developing recommendations for JPEG AIC (Advanced Image Coding).