In order to generate highquality radar echo images in complex environments and reduce the phenomenon of target information loss and relative imaging error, a radar echo image generation method based on small and mediumscale strong convection weather is proposed. The Fourier interpolation method, which can retain the radar echo characteristics to the maximum extent, is used to form the 3D regular grid data of radar echo with uniform spatial resolution. Based on the improved variational Bayesian theory, the radar echo image of small-and-medium-scale convective weather radar is super-resolved and generated. The experimental results show that the radar echo range of the medium-small scale strong convection weather radar processed by this method is highly similar to the original image. The radar echo information can be retained to the maximum extent. This method can generate clear radar echo images; there is no lack of target information; and the relative imaging error is very small.