Abstract: The micro-channel plate (MCP) imaging detector is a main detector used widely in detection of weak signals such as planetary atmospheric airglow and aurora. Detected from the unstable condition of outer space, the desired signal of MCP detector is mainly corrupted by Gaussian and non-Gaussian noise, which results in signal-to-noise ratio decrease and imaging quality deterioration. Due to the conventional filtering methods based on analog circuits can hardly control both the noise reduction effects of the signal and the count rate of the system at the same time, and cannot filter out the noise introduced by the post-stage circuit, a digital filter on the basis of a cascade structure was proposed for noise reduction of the observed signal. The method was arranged to carry out primary filtering firstly for the detected signals by a recursive least squares (RLS) filter, and then to input the filtered signals into a moving average (MA) filter for further smoothing, so as to get a high signal-to-noise ratio output signal. In order to test the noise reduction effect of the proposed RLS-MA filter, a comparison study was conducted among the least mean square (LMS), RLS, MA and RLS-MA filters with different signal-to-noise ratio signal inputs. The results show that the output signal of the RLS-MA filter is provided with higher signal-to-noise ratio and lower root mean square error.