Moving target classification method, system and device based on time-varying domain three-feature combination and medium

The invention discloses a moving target classification method, system and device based on time-varying domain three-feature combination and a medium, belongs to the technical field of signal classification, and solves the problems that a large amount of data learning is needed for feature analysis and processing in the conventional method for classifying by utilizing feature value extraction, and redundant features exist. The method comprises the following steps: firstly, receiving a motion sound source signal, and segmenting at equal time intervals; fourier transform is carried out on each small segment of the segmented signal to extract a line spectrum signal; transforming each small section of line spectrum signal to obtain three characteristic values, establishing a three-dimensional characteristic space by using the three characteristic values, and analyzing the distribution condition of two signals with moving target signals and background noise only so as to realize the classification of weak target si