tion of the original input signal because of the introduction of recent work has been devoted to the problem of joint coding and quantization noise. In fact, linear filtering cannot take into account decoding. In particular, to our knowledge, the design of decoders the nonlinearity of the quantization operation or the nonstationarhas been little investigated. In this article, we focus on decoding for ity of the images. Consequently, efficient coding necessitates a intraframe images of video data. In our approach, we propose a new global joint optimization of the coding/decoding chain.
method for decoding by nonlinear filtering. Here, we break with the This new method avoids the artifacts introduced by quantizausual approach of perfect reconstruction filters at the decoder, and tion (such as ringing) which can make prediction used in Pinstead pose the reconstruction model as a minimization problem. The frames more difficult. The proposed approach involves a joint solution can be viewed as an inverse problem with the optimization optimization of the transform/quantization/decoding structure.
of the transform/quantization/decoding structure formulated using a variational approach. We introduce sufficient conditions on the design
The solution of this problem can be viewed as an inverse problem of the decoder involving a priori assumptions on the solution and formulated using a variational approach. In other words, for a knowledge of the coder (transformation and quantization). We degiven analysis operator (filter) R and quantizer Q, we optimize velop an optimization method for the reconstruction filters at the dethe rate-distortion tradeoff by computing a nonlinear reconstruccoder to account for effects due to quantization noise. Experiments tion operator (filter) which minimizes a given criterion involving using this nonlinear inverse dynamic filtering demonstrate peak sigthe statistics of the noise subject to constraints. The constraints nal-to-noise ratio gains over standard linear inverse filtering as well are imposed using a priori assumptions on the solution, i.e., a as appreciable visual improvements. α§ 1998 John Wiley & Sons, Inc. Int priori knowledge about the image to be decoded. The criterion