On Image Charaterization in Real-Time

Abstract We present a machine model for the characterization of images that have the property of linear decomposition into strings of pixel-information. In this work characterization of images means recognition and modification of their graphical structures. Two types of the machine, the recognizer and modifier model, process images according to a programmed control unit with working tapes that adhere to structural information. Deprived of autonomous actions these machines are found to operate in real-time, serving as models for all practical implementations. It is shown via proper simulations that the real-time recognition property is preserved under most common image operations. The inverse image mapping obtained through a modifier is also characterized by a properly constructed recognizer machine.