Abstract: This paper analyzes the inclusions in the mud pulse signal noise. Correlation de-noising algorithm and moving median estimate algorithm to eliminate the mud pulse signal noise and pump base value drift were applied. Then for Manchester encode mud pulse signal characteristics, the correlation coefficient identification algorithm and local feature recognition algorithm were applied to the identification of the mud pulse signal. The experimental results show that the process of de-noising and decoding algorithm is simple and practical, high reliability, low decode error rate. All these comply with the requirements of the engineering application.