In this study, the notion of Remote Sensing Intelligent Geo Interpretation Model (RSIGIM) is firstly presented. The main task of RSIGIM is to automatically simulate the human’s interpretation to the remote sensing image. The initial target of RSIGIM is to describe, recognize, extract, classify and interpret the feature from RS image, and further to identify Geo properties and inner characters of the feature, such as size, structure, relationship. Integrated with geographical analysis model, the end target aims to mine out the geographical phenomenon, to predict developing process hidden in the large capacity of spatial data. The crucial issue of RSIGIM is to establish its cognition model, which is initially designed as a hierarchical structure. From low layer to high layer, three components, including statistical processing and analyzing model, neural computation and evolutionary computation based vision physiological cognition model, symbolic knowledge logic reasoning based vision psychological cognition model, are integrated to constitute relatively perfect cognition model.