The team of Mohammed Aabed, Gukyeong Kwon, and Ghassan AlRegib was named as a finalist for the First 10K Best Paper Award at the 2017 IEEE International Conference on Multimedia and Expo, held July 10-14 in Hong Kong. They were recognized for their paper entitled “Power of Tempospatially Unified Spectral Density for Perceptual Video Quality Assessment.”

A professor in the Georgia Tech School of Electrical and Computer Engineering (ECE), AlRegib leads the Multimedia and Sensors Lab, where Kwon is a current Ph.D. student in ECE. Aabed graduated with his Ph.D. in ECE in fall 2016 under the direction of AlRegib, and he now works at Amazon Lab126 in Sunnyvale, California.

The objective of the research presented in this paper is to investigate how the human visual system perceives quality of scenes, videos, and images. The novelty of this research lies in revealing the relationship between power spectral density (PSD) and perceptual video quality. This team's work shows the efficacy of statistical features from 3D PSD in capturing the impact of different distortion types, distortion levels, and scene characteristics on videos.

Besides its effectiveness in characterizing visual scenes, this approach is significantly computationally inexpensive. This novel representation of videos can be used to solve a variety of vision problems in practical applications beyond understanding the way humans perceive quality. For example, the team can foresee utilizing this research in a broad spectrum of applications such as object classification and detection, and scene understanding for autonomous vehicles.