The invention provides a gesture recognition method based on 77GHz millimeter wave radar signals. The method comprises the following steps: acquiring intermediate frequency signals of different gesture actions through radar; innovatively utilizing an improved wavelet threshold function to preprocess a low-frequency coefficient of the wavelet threshold function so as to solve the problem that a close-range gesture cannot be recognized due to an antenna coupling phenomenon; secondly, extracting a time-distance spectrogram, a time-speed spectrogram and a time-angle spectrogram from the preprocessed intermediate frequency signal; innovatively splicing the three characteristic spectrograms to obtain the diversified characteristic diagram, and inputting the diversified characteristic diagram into the convolutional neural network for training. The problem of incomplete information expression of a traditional identification algorithm is optimized, the network structure is simplified, and a relatively good identification effect is finally obtained.