The asymptotic closed-loop approach to predictive vector quantizer design with application in video coding

Abstract

The basic vector quantization (VQ) technique employed in video coding belongs to the category of predictive vector quantization (PVQ), as it involves quantization of the (motion compensated) frame prediction error. It is well known that the design of PVQ suffers from fundamental difficulties, due to the prediction loop, which have an impact on the convergence and the stability of the design procedure. We propose an approach to PVQ design that enjoys the stability of open-loop design while it ensures ultimate optimization of the closed-loop system. The method is derived for general predictive quantization, and we demonstrate it on video compression at low bit rates, where it provides substantial improvement over standard open and closed loop design techniques. Further, the approach outperforms standard DCT-based video coding.

Challenges and Recent Advances in Distributed Predictive Coding

“…The asymptotic closed-loop (ACL) design approach [20], [19] overcomes these shortcomings of traditional predictive coder design. A subterfuge is employed wherein the design is effectively performed in open-loop, where each quantizer is designed for the statistics of the exact signal it then quantizes to produce a new sequence of reconstruction for the next iteration, thereby circumventing stability issues.…”

“…Asymptotically, the loop is effectively closed in the sense that the design approaches closed-loop statistics despite open-loop operation within each iteration. For a detailed treatment of ACL and its applications, see [19], [20].…”

“…This will be explained in more details in Sec. II-E To circumvent these difficulties we use the technique of asymptotic closed loop (ACL) { [19], [20]} which we rederive for DPC system design. Within the ACL framework, the design is effectively in open-loop within iterations (eliminating issues of error buildup through the prediction loop), while ensuring that asymptotically, the prediction error statistics converge to closed loop statistics.…”

Predictive multistage vector quantizer design using asymptotic closed-loop optimization

“…We previously proposed a solution to this problem in [1] and [5], for the simpler case of single-stage predictive vector quantizers (PVQ), which offers several attractive features, and achieves significant gains over traditional design methods. Moreover, the coding performance was competitive with H.263.…”

“…In this correspondence, we also propose a selective splitting codebook design algorithm that grows codebooks gradually and optimally. The proposed selective splitting design is applied to variable rate multistage vector quantizers, and is found to be effective when incorporated into the predictive quantizer design approach [1]. Simulation results for video coding are provided.…”