Abstract: One of the important tasks of the post genome project is protein structure prediction，the key step of which is protein secondary structure prediction.This paper makes use of a model of digital coding for amino acid，and uses cellular automata to generate image representation for protein sequences.A protein sequence can be represented by a unique image，and the image takes into account the interactional actions between amino acids.By the Gray Level Co-occurrence Matrix（GLCM） deriving from the CAI，a novel method has been developed that can be used to predict the secondary structure of the protein.The result thus obtained is quite promising.