The invention discloses a two-dimensional cellular automaton evolution rule classification method and belongs to the technical field of a cellular automaton. According to the method, under the random initial conditions, statistics of change conditions of cellular states of each position in the certain-scale evolution space is carried out, and classification is carried out; if within plenty of continuous time steps, the evolution space at each time and the stable evolution space have non-big difference, and the evolution space is a stationary type; if within plenty of continuous time steps, one period exists, the evolution space at each time in the period and the evolution space at the time of the next period have non-big difference, the evolution space is a period type; if within plenty of continuous time steps, the evolution space at each time is complex, the evolution space at one time has a relatively large quantity of cellulars in an update state compared with the evolution space at the previous time, update is irregular, the evolution space is a complex type; otherwise, the evolution space is a confused type. The method is advantaged in that the method can be applied to image encryption and image compression, application evolution rules are wide, and accurate classification can be realized.