基于彩色相机的水下物体表面光谱反射率重建方法研究

Light attenuation by water has severe impact on underwater color imaging. Investigating the physical principles on underwater color imaging and reconstructing the spectral reflectance of underwater objects accurately is very important for automatic detection and recognition of underwater objects, color reproduction of underwater environment, etc. This project focuses on spectral reflectance reconstruction for underwater objects based on color camera. Theoretical model will be built for the underwater color imaging process, based on the theory on optical imaging, underwater light propagation, color, etc. Physical insight will be gained on the underwater color imaging process and the principles behind will be revealed. Based on the theoretical model, a method will be proposed to accurately calibrate the spectral attenuation coefficient of water column by making use of color images, where the spectral attenuation coefficient will be used to quantify the water attenuation and get rid of the influence of water on spectral reflectance reconstruction. To achieve the mapping from RGB to spectral space,the water column and underwater object will be considered as a virtual object in air and a model will be identified by numerical fitting to map the RGB value of the virtual object to its spectral reflectance. After spectral reflectance is reconstructed for the virtual object, the influence of water will be removed in spectral space based on the water spectral attenuation coefficient, and the spectral reflectance of underwater object is obtained. This project is defined closely related to the national strategy on ocean development. Research results will serve as both theoretical and experimental basis for applications like automatic detection and classification of underwater objects and color reproduction of underwater environment, thus contributing to the development of color theory and its application in the oceans.

水体对光的衰减对水下彩色成像造成严重影响，探索水下彩色成像物理规律、并对水下物体表面光谱反射率进行精确重建，对于水下目标的探测和识别、水下物体颜色的真实再现极其重要。本项目拟针对“基于彩色相机重建水下物体表面光谱反射率”这一问题展开深入研究。以相关理论为基础，建立水下彩色成像理论模型，揭示水下彩色成像的物理规律；在理论模型的基础上，提出基于彩色图像对水体光谱衰减系数进行精确定标的方法，用于后续量化和去除水体衰减；提出将水体和水下物体视为一个虚拟整体的思路，采用数据拟合的方法，建立从该虚拟物体的彩色图像到其光谱反射率的映射模型；结合水体衰减系数，在光谱空间去除水体光谱衰减，得到水下物体真实的光谱反射率。本项目的制定紧密结合国家的海洋战略，研究成果将为水下目标的自动探测和识别、水下环境颜色的真实再现等提供理论和实验基础，为颜色科学理论的发展及其在海洋领域的应用做出贡献。

光谱成像技术可以同时获取目标物在空间维度和光谱维度的信息，因此它在水下目标探测和识别、海洋生态监测、水下环境颜色的真实再现等领域有着重要作用。.本课题从理论上分析了光的水下传播过程，包括光在水体中传输的过程中的吸收、散射等效应；设计并搭建了水下光谱成像系统；设计并实现了水下目标物表面光谱反射率重建算法，最终获取了水下目标物的光谱特性。算法中分别对光在水体中的散射、折射、衰减和系统测量误差进行了修正补偿，最终通过算法补偿获得了准确的水下目标物表面光谱反射率信息，还原了水下目标物的真实颜色，为光谱图像和颜色识别在水下应用提供参考。