Abstract: A video image decoder architecture for implementing a processing algorithm in the 40-ms mode on high-resolution TV sets, of a kind adapted to handle TV signals being received on respective transmission channels (J,L), which comprises a video signal demultiplexer receiving the transmission channels (J,L); and respective processing blocks for separately handling the signals from each of the channels (J,L). Each processing block includes a video image format converter, a local memory connected to an output of the converter, and at least one median filter and one systolic filter cascade connected after the memory for restoring, by interpolation, signal samples related to successive lines of the video image. A summing node adds the outputs from each processing block so as to obtain a time mean between restored samples of the channels (J,L).