Cellular automata produce spatial patterns when specific rules for time development are given. This paper deals with an inverse problem of identifying the rules for spatial patterns given. Although only rules of one-dimensional elementary cellular automata and one-dimensional probabilistic cellular automata have shown here, the system can deal with two-dimensional one. When the rule identification has not been fully successful due to the lack of information in the spatial pattern, the system is able to give an identifiable part of the rules with a format of Wolfram’s rule number.