This paper addresses the problem that the detector α-AMF is not fully adaptive and the control parameter α is mismatched when using the sample covariance matrix(SCM) estimation in the compound-Gaussian background. Firstly, combined with the normalized sample covariance matrix(NSCM) estimation, a SCM-NSCM combination estimation method for the α-AMF is proposed. Then, the empirical formula of the optimal control parameter is simulated and it is consistent with the numerical results. Finally, the constant false alarm rate(CFAR) characteristics and the performance of the detectors are analyzed. The results show that the α-AMF based on SCM-NSCM estimation, which is less affected by the spikiness of clutter than the comparative detector, has strong robustness to the variation of the normalized covariance matrix of the clutters in the compound-Gaussian environment, and it exhibits better detection performance than the comparative detector in heavy-tailed clutter.