Evaluation of images by means of recognizability, which is independent of the techniques for physical image evaluation so far being used, is proposed. This evaluation method can he defined by extending one-dimensional sequential dependency in information theory to two-dimensional one. The present paper defines the strongness of sequential dependency and the magnitude of recognizability. By making use of this recognizability, quantitative evaluation of those images which may not be rationally evaluated by the conventional physical evaluations, such as modulation transfer function and Wiener spectrum, became possible to he made.
抄録全体を表示