Full-Stokes parameters of reflected lights provide rich information on a target’s polarization properties, making polarimetric imaging play a crucial role in numerous application fields of optic vision, where vast amounts of information are required. However, efficiently and accurately retrieving the full-Stokes parameter remains a challenging task, as it usually involves a complex process of polarization calibration. In this paper, we propose what we believe to be a novel method to directly retrieve the full-Stokes parameter for snapshot polarimetric imaging by introducing an instrument matrix of the imaging polarimeter for polarization calibration. Specifically, a series of spatially-homogeneous polarized lights with determined Stokes vectors are firstly generated to inject the polarimeter to experience a multi-channel polarization modulation, and for each injection the incident light field is captured to yield an intensity vector accordingly; then based on those input Stokes vectors and the output intensity vectors, an unconstrained optimization problem is proposed and solved by introducing a regularization term related to the degree of polarization, which gives the instrument matrix of the polarimeter; finally the full-Stokes parameter of a scene for practical polarimetric imaging can be retrieved directly by multiplying the obtained instrument matrix and the intensity vector for each snapshot. The experiments carried out with a self-developed division-of-aperture polarization camera verified the effectiveness of the proposed method, with a maximum absolute error of no more than 0.031, and an order of magnitude reduction of the root mean square error.