An optical coherence tomography system will inevitably introduce additional aberration into scanning imaging. Thus, image details are lost, and the high definition requirements of medical imaging cannot be satisfied. In this study, we proposed an aberration correction method based on the particle swarm optimization algorithm. The aberration correction method was modeled as a filtering process based on a linear combination of the Zernike polynomials. The particle swarm optimization algorithm was used to iteratively estimate the optimal coefficient value of the polynomials by considering the image information entropy or sharpness as the optimization index. Further, the resolution plate was used as the simulation target image to load the defocused and low-order mixed wavefront aberrations. The image information entropy and sharpness were considered to be the evaluation functions. The root mean square error values of the restoration results were less than 0.1λ, so the clear imaging can be obtained. For the experimentally collected onion cells, the information entropy was considered to be the evaluation index, which decreased by 18% after aberration correction. However, for the collected grape tissue image, sharpness was considered to be the evaluation index, which increased by 36% after aberration correction.