为了改善可见光图像质量, 提高对隐藏在自然背景中人造目标的探测率, 采用了基于偏振成像的可见光图像增强方法, 很好地凸显出人造目标, 提供更多的细节与纹理信息。利用偏振成像方式获取偏振度、偏振角和椭率角图像, 对多维偏振图像中的像素点在邦加球内进行聚类得到偏振特征图像, 运用双树复小波对光强图像和偏振特征图像进行分解, 低频子带采用主成分分析法, 高频子带选用区域方差特征匹配的融合策略, 得到增强图像。进行了户外实验, 并理论分析得到了各个图像的性能指标数据。结果表明,增强后的图像较原图像具有更加丰富的图像细节与偏振目标信息, 该方法有效可行,对目标识别与探测具有一定意义。

In order to improve the image quality and recognition of an artificial target hidden in the natural background, an enhancement method of the visible light image based on the polarization imaging was put forward to highlight the artificial target and provide more details and texture information. Firstly, the degree of polarization, polarization angle and ellipticity angle of the image were obtained based on polarization imaging. Secondly, the polarization characteristics were obtained after the pixels in the multidimensional image clustered in the Poincare sphere. Fianlly, the intensity image and the polarization characteristic image was decomposed by means of the dual tree complex wavelet transform. The principal component analysis was used in the low frequency sub-band and the feature fusion strategy based on the region variance was used in the high frequency sub-band, then the enhancement image was obtained. The outdoor experiment was performed to test the effectiveness of the algorithm proposed in the paper. After the theoretical analysis, the performance index of each image was obtained. The results of the subjective and objective evaluation both show that the enhancement image has more details and polarization information, which are important for target recognition and detection.