Abstract: Considering the shortcoming that the local binary patterns (LBP) is easily effected by noise, a novel improved local ternary patterns (rotation invariant LTP or RILTP) is used to estimate the camouflage images in this research. Instead of a single operator in LBP, the LTP uses positive and negative mode operators to describe image texture. As a result, the describing accuracy and noise immunity is improved greatly. Based on subjective assessment of the human eye, RILTP has been compared experimentally with the traditional LBP to verify its validation. The results show that RILTP has better consistency with subjective assessment than classical LTP.