Digital holographic microscopy (DHM) is a method for reconstructing the three-dimensional (3D) profile of an object. However, the reconstructed 3D profile often contains the noise caused by the direct current spectrum. Conventional methods struggle to extract only the phase difference information from Fourier domain, leading to the containing the noise spectrum. To solve this problem, noise reduction methods using the Kalman filter have been proposed. However, conventional methods use fixed filter parameters that cannot adapt to time-varying noise characteristics. In this research, we propose an adaptive filtering method that dynamically adapts filter parameters based on the error magnitude. This approach enhances the noise suppression as the system stabilizes, enabling the reconstruction of the 3D profile under the noise-reduced. To evaluate the effectiveness of our proposed method, we conduct a comparative analysis using numerical evaluation metrics. The results demonstrate that our proposed method improves the noise suppression performance while maintaining the computational efficiency, making the adaptive filtering feasible.