In order to improve the poor quality and low contrast of underwater polarization image, an underwater polarization image fusion algorithm based on Retniex and wavelet transform is proposed. Firstly, the intensity image was preprocessed by Retniex enhancement. Then the intensity image and polarization image were decomposed by wavelet transform (DWT). The low frequency component was fused by a fusion rule based on significance graph, and the high frequency component was fused by increasing the absolute value of pixel. Finally, the inverse wavelet transform was used to obtain the fused enhanced image. The experimental results show that the visual effect of images achieved by this algorithm is better, with more obvious details and higher clarity. Objectively, the information entropy, contrast and average gradient of fused images are improved by 6.6%, 98.5% and 39.2% respectively, and the quality of underwater images is significantly improved.