Ricardo de Queiroz
Universidade de Brasília (Brazil)
Video CODEC evaluation through rate-distortion-complexity analysis
Language: Portuguese
Date: October 8th, 10h00
Abstract
We study the decision process for choosing video compression systems through a rate, distortion, and complexity (RDC) analysis. Bjontegaard delta (BD) metrics are generalized and expanded into the 3D RDC space. We discuss metrics to evaluate a CODEC characterized by a cloud of points or a line of points in RDC space. Our presentation will focus on a Lagrangian cost D + λR + γC , so that the (λ,γ) choice depends on the application. We define, then, application spaces and examples will be given of applications where (λ,γ) are calculated, such that we can objectively compare Lagrangian costs in an RDC volume for different CODECs for a given application. Furthermore, we can span the plane and compare CODECs for the entire application space filled with different (λ,γ) choices. We compared several state-of-the-art neural video CODECs using the proposed metrics and found that only four neural video CODECs came out as the best suited for any application.
Biography
Ricardo de Queiroz is a graduate from Universidade de Brasília (1987), obtained his Master’s from Universidade Estadual de Campinas (1990) and his Ph.D. from The University of Texas at Arlington (1994), all in Electrical Engineering. He is a Full Professor at Universidade de Brasilia. He has large experience in digital images and video, as well as in signal processing. Prof. De Queiroz is a Fellow of IEEE, a Researcher 1A at CNPq, and, according to Google Scholar (R), his work has been cited about 7400 times with an h-index of 47.