The invention discloses a method and system for reconstructing incoherent motion magnetic resonance imaging parameters in voxels. The method comprises the following steps: setting a parameter D, a parameter f, a parameter D * and a parameter S (0) in a geometric figure generated in a simulation area, and judging whether the total area of all geometric figures covers the simulation area or not; ifso, generating a D parameter graph, an f parameter graph, a D * parameter graph and an S (0) parameter graph; generating a magnetic resonance diffusion weighted image corresponding to each b value, and training the neural network model to obtain a trained neural network model; and performing Fourier transform and normalization processing on the k-space data, and inputting the normalized magnetic resonance diffusion weighted image into the trained neural network model to obtain a reconstructed IVIM parameter image. By adopting the method and the system provided by the invention, the problem that the reconstruction result presents granular sensation due to point-by-point fitting is solved, the image is smoother, the influence of the small b value on the IVIM double-exponential model is considered, and the reconstruction effect is improved.