The light field Micro-PIV deconvolutes the light field image by Lucy-Richardson (L-R) algorithm to obtain the three-dimensional coordinate information of the tracer particle. The accurate point spread function (PSF) is the premise of the L-R algorithm to complete the reconstruction, and the existing PSF model is not suitable for the light field Micro-PIV system. In this paper, the PSF model of a microscopic light field imaging system based on wave optics is established. Numerical simulation is carried out to calculate the PSF. The similarity between the calculated and experimental PSF images is determined by the structural similarity algorithm. Combining the theoretically calculated PSF, the L-R algorithm is used to reconstruct the single particle and the flow field of tracer particles at different concentrations. Results show that the similarity between simulated and actual PSF is more than 0.94, indicating that the PSF model has high accuracy. The three-dimensional coordinate error of a single particle is within one pixel, and the three-dimensional coordinate information of the tracer particles at different concentrations can be accurately obtained, which further verifies the accuracy of the PSF model and lays the foundation for light field Micro-PIV to realize instantaneous three-dimensional velocity field measurement.