Abstract: Photogrammetric image matching/correlation is an important aspect of computer vision/machine vision, Many image matching algorithms make use of the local gray information only and utilize a single criterion for matching processing, that is one of the reasons why the reliability of the results is usually very low, In this paper, information theory, human-visual-system research achievements and photogrammetric stereo-observation experiences are introduced to analyse the factors influencing the matching reliability. Based on that, the theory of multi-information and multi-criterion image matching comes thus into being. The method of image matching aided by image structural characteristics and the adaptative control tcchnique are briefly described.