Aberration is a crucial factor that restricts the imaging performance of optical systems， leading to degraded imaging effects such as blurred details and reduced resolution. Computational optical aberration correction technology breaks through the limitations of traditional hardware-based correction by integrating optical physical modeling with advanced information processing algorithms， achieving accurate and flexible compensation for imaging degradation caused by various aberrations. This paper systematically reviews the research progress of information processing methods for aberration correction based on computational optical systems. First， it expounds the theoretical foundation of aberration correction， including the wavefront aberration characterization method based on Zernike polynomials， which provides a rigorous mathematical basis for quantifying different types of aberrations， and the aberration-dominated light field degradation model that establishes the quantitative correlation between wavefront distortion and point spread function （PSF） degradation. It also introduces classical restoration algorithms such as Wiener filtering and Richardson-Lucy iteration， which lay a technical foundation for solving the imaging degradation inverse problem. On this basis， the paper analyzes the principles and practical applications of three mainstream aberration correction technologies from the perspectives of active adjustment， optical coding and pure computational restoration： Adaptive optics （AO） realizes real-time dynamic compensation of dynamic wavefront distortion through a closed-loop system of wavefront sensing and deformable mirror adjustment； wavefront coding and coded aperture technology transform complex aberrations into computable degradation forms via artificial phase modulation in the optical front-end， realizing collaborative optimization of optical coding and back-end digital decoding； phase retrieval and blind deconvolution techniques invert wavefront phase information and estimate unknown PSF only through algorithm iteration without additional hardware intervention. Finally， the paper focuses on deep learning-driven aberration correction methods， including data-driven end-to-end learning frameworks， physical model-embedded hybrid architectures and unsupervised/few-shot learning methods， and discusses their typical applications in biomedical microscopic imaging， lensless imaging， astronomical remote sensing and other frontier fields. This study clarifies the technical characteristics， advantages and limitations of various methods， and provides important theoretical reference and technical path guidance for the collaborative optimization design and practical application of computational optical imaging systems.