空气污染严重影响了成像质量, 雾霾天气下成像的对比度和分辨率下降。 利用不同波段偏振光谱数据的图像去雾研究国内外尚未见报道； 运用偏振光谱去雾技术对光谱数据立方体研究具有重要意义。 在中重度雾霾天气下, 利用光谱仪, 获取了380～1 000 nm的可见近红外偏振光谱数据, 研究了450, 550, 650, 750和850 nm五个波长下偏振图像和去雾结果。 结果表明, 波长对图像目标与背景的灰度差值影响显著, 去雾后图像的灰度动态范围和平滑度都有显著提升； 去雾后图像远景对比度平均提升了5倍, 其中450 nm图像对比度最低, 但对比度提升量最大, 提高了7.13倍； 850 nm图像对比度最高, 但对比度提升量最小, 提高了3.86倍； 去雾后不同波长图像的灰度动态范围均有明显展宽, 展宽范围为13.5%～28.6%, 而图像近景动态范围展宽为33.3%～44.0%。 分析了图像复原过程中可能出现的估计误差, 根据其产生机理提出了对大气散射光的偏振度和无穷远处的大气散射光强度的校正方法, 并总结出了最佳校正系数的选取规律, 对未知条件下的图像去雾提出指导。 通过斯托克斯参数从原始数据中提取偏振信息, 基于大气散射光与景物透射光偏振特性差异对图像去雾, 偏振光谱图像去雾技术为光谱图像去雾研究提供一种新手段, 也为图像去雾技术的发展提供一种新思路。

Air pollution has a serious effect on the quality of image, and image taking under hazy weather suffers from poor contrast and resolution. It is of great significance to use the polarimetric spectral information for image dehazing and the research of polarization dehazing for spectral data cube. The data used in this paper was pushbroomed under moderate hazy weather, and original Vis-NIR polarimetric data of 380~1 000 nm was captured, the original polarimetric images at 450, 550, 650, 750, 850 nm and the dehazed images were studied. The results of the research showed that wavelength had a significant effect on the difference of gray value between targets and background, and both the dynamic range of gray value and the smoothness of histogram were improved after the dehazing process. On average, the image contrast of the far-field targets increased by 5 times. The far-field targets at 450 nm had the lowest contrast, and the dehazing process increased the image contrast by 7.13 times which made the undetectable targets detectable. The far-field targets at 850 nm had the highest contrast, and the dehazing process increased the image contrast by 3.86 times. The dynamic range of histogram of the full image increased to 13.5% to 28.6% from 450 to 850 nm, the dynamic range of histogram of the near-field targets are increased to 33.3% to 44.0%. Based on the analyzation of the possible estimation error, two correction factors were proposed to revise the degree of polarization of the airlight and the intensity of the airlight from an object at an infinite distance, the regularity of the two factors were given to guide the image dehazing process under new conditions. The polarimetric information is obtained through the original data by the Stokes parameters, the dehazing process is based on the polarimetric difference between the air light scattered from the haze particles and the direct light reflected from the objects. The polarization spectral dehazing technique not only expands the application area of the imaging spectrometer, but also provides a new idea for image dehazing.