Abstract: The estimation error of reconstruction by adaptive compressed sensing was derived, and the column vector autocorrelation of the observation matrix was reduced, and the impact of optimization process on compressed sensing reconstruction algorithm was analyzed. Combining the observation matrix optimization and adaptive process, the spectrum sensing algorithm of optimized adaptive compression based on observation matrix was proposed. The simulation results show that the mean square error (MSE) of proposed algorithm is lower than traditional algorithm, and the probability of detection of proposed algorithm is higher on the same number of observations, and the required number of observations is fewer when achieving the same receiver operating performance (ROC).