为了改善了水下偏振图像存在的质量差、纹理细节模糊和对比度低等问题。提出基于区域方差的水下偏振图像融合算法。首先, 利用532 nm激光器拍摄不同角度的水下偏振强度图, 计算相应的斯托克斯图像; 然后, 对Dolp-Aop图像进行同态滤波处理, 将其和光强图像I进行三层小波分解(DWT); 接着, 将小波分解源图像产生的低频子带和高频子带分别采用不同的融合算法实现融合。仿真结果证明, 本算法实现的融合图像与传统的融合算法相比, 融合图像在对比度和清晰度上有明显提高。

In order to improve the poor quality of underwater polarization image, blurred texture details and low contrast. An underwater polarization image fusion algorithm based on regional variance is proposed. Firstly, the 532 nm laser is used to record the underwater polarization intensity maps at different angles, and the corresponding Stokes image is calculated. Then, the homomorphic filter processing is performed on the Dolp-Aop image, and the light intensity image I is subjected to three-layer wavelet decomposition (DWT). Next, the low frequency sub-band and high frequency sub-band generated by the wavelet decomposition source image are respectively combined by different fusion algorithms. The simulation results show that compared with the traditional fusion algorithm, the fused image realized by this algorithm has a significant improvement in contrast and sharpness.