A Model for the Random Video Process

This paper provides a detailed development of a simple model for characterizing the statistical properties of a random video signal. The primary concern is the modeling of the power spectral density of the electrical process generated by linear, sequential scanning of a rectangular portion of an infinite, two-dimensional random picture. The spatial and temporal statistical properties of typical picture material allow approximations which lead to a model having an especially simple form, characterized by only a few parameters. The model has a form convenient for the analysis of a variety of signal transmission problems. The validity of the model for these purposes is established by comparing it with results obtained in several independent experimental studies. 12,3 The relationship between the second-order statistics of the random picture and those of the resulting video signal due to line-to-line and 609 610 T H E B E L L S Y S T E M TECHNICAL JOURNAL, A P R I L 1966 frame-to-frame correlation is examined in Section II. Section III considers the composite video signal wherein the picture signal is periodically interrupted and a periodic pattern is inserted for purposes of synchronization and blanking. A model of the random picture process is developed in Section IV. The results are combined in Section V to provide a summary of expressions for the power spectral density of the composite video signal. Section VI is an illustrative example of the use of the model for deriving optimum linear signal processing networks for video signal transmission over a noisy channel.