雾霾降质图像全偏振清晰化方法及探测器研究

The low visibility caused by the fog and haze has a direct impact on the normal operation of the vision system in the fields of military reconnaissance, civil aviation transportation and others. According to the theory of atmospheric physics, applying the full polarization property of light can improve the effect of the visual perception in low visibility conditions. This project focuses on solving two key scientific problems, including the influence mechanism of fog and haze to the full polarization property and the fog and haze imaging retrieval model, in order to progressively research the fog and haze degraded image clearness method and detector based on full polarization. Initiatively, the materials polarizing properties and the full polarization transmission properties are deeply clarified under fog and haze environment with the various categories, multiple scales and space-time unsteady distribution. Next, the non-linear mapping rules between the full polarization parameters and imaging parameters are mainly explored to build the fog and haze imaging retrieval model. Once again, the four-channel space-time synchronous full polarization detector with image stabilization is developed by proposing the linkage compensation compliant coordination control strategy and achieving the method of the space-time synchronous normalization to four-channel full polarization information. Finally, on the basis of the tests on the UAV experiment platform under the different degrees of fog and haze conditions, the performances of the method and the detector mentioned above are respectively evaluated by the visible edge contrast enhancement evaluation method and non-uniform sampling error analysis model. The project tightly combines the concept and method of atmosphere physics and the practical engineering, in order to supply an innovative research approach to solving the problem of the fog and haze degraded image clearness.

雾霾导致能见度降低，直接影响军事侦察、民航运输等领域中视觉系统的正常运行。根据大气物理学理论，利用光的全偏振特性能够提高低能见度条件下的视觉感知效果。本项目着重解决雾霾对全偏振特性影响机理和雾霾成像反演模型两大关键科学问题，逐层递接地开展雾霾降质图像全偏振清晰化方法及探测器研究。首先，阐明多种类、多尺度、时空非定常分布的雾霾环境下物质起偏特性和全偏振信息传输特性；其次，探索全偏振信息参数与雾霾成像参数间的非线性映射规律，建立雾霾成像反演模型；再次，提出联动补偿柔顺协调控制策略，实现四通道全偏振信息时空同步归一化校正方法，研发四通道时空同步稳像全偏振探测器；最后，基于无人机试验平台完成不同程度雾霾条件下的测试，采用可见边对比度增强评估方法和非均匀采样误差分析模型分别评价所提出的方法及探测器性能。本项目将大气物理学概念、方法与工程实际紧密结合，为解决雾霾降质图像清晰化难题提供一种新的研究思路。

因工业污染、全球变暖等因素，雾霾天气频繁出现，覆盖范围逐年扩大，据环保部通报数据显示，我国雾霾影响面积高达143万平方公里，共涵盖25个省份。雾霾是降低视觉感知有效性的主要因素，直接影响到军事侦察、民航运输及港口物流等领域中等视觉系统的正常运行。本项目组着重解决雾霾对全偏振特性影响机理和雾霾成像反演模型两大关键科学问题，逐层递接地开展雾霾降质图像全偏振清晰化方法及探测器研究。首先，阐明了雾霾环境下物质起偏特性和全偏振信息传输特性；其次，以雾霾对全偏振特性影响机理为理论依据，探索了全偏振信息参数与雾霾成像参数之间的非线性映射规律，建立了雾霾成像反演模型；再次，基于项目组前期对线偏振信息获取装置的概念设计，研究了四通道全偏振探测器机构的实现方案，提出了用于稳像功能的联动补偿柔顺协调控制策略，设计了四通道时空同步稳像全偏振探测器，同时出于对配准精度的考虑研制了单通道快速旋转式偏振探测器；最后，利用本项目组所制全偏振探测器及所提雾霾降质图像清晰化算法对地面雾天景物进行反演处理，并采用信息熵、平均梯度及灰度方差三个指标客观评价清晰化效果。