The property of optoelectronic detector indicates that the response between incident light intensity and output digital number is non-linear. For imaging polarimeter, if light intensity is directly substituted by digital number when restoring polarization information from polarization images, the restored polarization information will apparently deviate from the true scene polarization information. This deviation makes the quantitative application of polarization information meaningless. To solve this problem, in this paper we propose a new algorithm for restoring the polarizaiton information with considering the property of polarimeter of the detector. Firstly, theorical correction equation of polarization information restoration are proposed for division-of-time polarimeter and for simultaneous polarimeter respectively. And then, specific implementation and polarization imaging test experiment are carried out. For divison-of-time polarimeter, we firstly test the property of the detector, and then use the tested parameter and the proposed restoration equation to restore polarizaiton information from the data of polarzation imaging test experiment. The degree of linear polarization (DoLP) restored with digital number directly changes from 0.932 to 0.753 when changes from 1.0 to 1.5. The DoLP restored with correction equation proposed in this paper varies from 0.932 to 0.926, which approaches to the ture scene DoLP value 1.0. For simultaneous polarimter, the instrument matrixes of the polarimeter are calibrated first under different setting values, and then the calibrated instrument matrixes are used to restore poliarzation information from the data of polarization imaging test experiment. The DoLP restored with digital number directly changes from 1.3763 to 1.1582 when changes from 1.0 to 1.5, which exceeds the possible DoLP range from 0 to 1.0. The DoLP restored with correction equation proposed in this paper varies from 0.8428 to 0.8683, which approaches to the ture scene DoLP value 1.0. Experimental result shows that the property of polarimeter has an apparent effect on the restored polarization information, and that the polarizaiton information restoration error increases with setting deviation from 1.0. With the restoration algorithm proposed in this paper, the restored polarization information can steadily approach to the scene polarization information with acceptable slants small. The poroposed polarization restoration algorithm with considering property establishes a theoretical foundation for the future study of polarimeter and its quantitative application.