This paper addresses the problem that the optimal or suboptimal clutter covariance matrix estimation method in the specific clutter background is difficult to adapt to the transitional clutter environment. Firstly, a fusion estimation method for the matrix covariance matrix structure, which covers three existing clutter covariance matrix estimation methods by adjusting the parameter, is proposed, and the adaptive characteristics of the adaptive normalized matched filter corresponding to the proposed method are analyzed and verified by simulation experiments. Then, the empirical formula of the parameter is determined, and it conforms to the numerical results. Finally, the proposed method and the existing methods are compared and analyzed from the aspects of estimation accuracy, constant false alarm rate and detection performance. The results show that the proposed method, which has strong ability to adapt to the temporal and spatial gradients of the clutter, has higher accuracy and better detection performance in the transitional clutter environment.