The invention discloses an adaptive CSI signal auxiliary filtering method based on frequency analysis. The method comprises the following steps of 1, collecting one segment of environment noise data before collecting required CSI data containing motion information; 2, collecting the CSI data containing the motion information and carrying out frequency domain analysis comparison on the two segmentsof data; and 3, determining the passband cutoff frequency and the stop band cutoff frequency of a low pass filter based on a frequency difference and filtering a signal. In the invention, the processing time of CSI signal filtering is shortened, and a support is provided for constructing a real-time system used for detecting human body behavior information.