In order to better monitor the underwater environment and seafood in marine pastures， an underwater polarization imaging system is integrated by combining a 532 nm laser light source with wave sheet modulation， and the polarization image is processed by Miller matrix method. The experiment takes sea cucumber as the target and simulates the water environment of Marine pasture. After obtaining the polarization image through the imaging system， image fusion is performed according to the derived Miller matrix formula. By comparing the contrast and information entropy between the fused Miller matrix image and the original image， it is proved that the Miller matrix method can effectively eliminate the influence of background scattered light to improve the imaging contrast and quality.