节点文献
基于光流约束方程的稀疏光场稠密重建方法研究
Research on Dense Reconstruction Method of Sparse Light Field Based on Optical Flow Constraint Equation
【作者】 刘健；
【作者基本信息】 中北大学 ， 数学， 2022， 硕士
【摘要】 光场以能够记录现实环境中的高维光线数据的特性来感知周围动态环境的变化,并以此发挥光场在智能系统中对环境决策的作用。光场成像技术突破了经典成像模型与数字相机的局限性,通过光场全光函数的表达方式,明确了光在时间域、空间域以及光谱等多个维度的联系,进而完成了耦合感知、解耦重建等智能处理的任务,同时使得光场能够对多维大范围动态场景进行成像。光场成像技术因其独特的优势在重聚焦、虚拟现实、生命科学和国家安全等领域得到了长足的应用。然而,伴随着高维数据的离散化采样,光场成像技术正面临着同时获取高空间分辨率与高角度分辨率光场的基础难题,如何对稀疏化角度维度的光场数据进行高质量的稠密重建成为亟待解决的问题。本文针对光场成像技术无法同时获取高空间分辨率与高角度分辨率光场的问题,开展了光流场的稀疏光场稠密重建算法的研究,本文从二维光流基本原理出发,以光场双平面模型为基础,提出一种新的四维光场光流约束方程数学模型,利用该约束方程求解得到的光场光流确定新视角的位置坐标,并通过插值计算及图像反演逐点获得新坐标的强度值,最终获得高质量的新视角合成图像。多个不同场景的实验验证了本文算法具有很强的适应性。同时,定量评价结果说明本文算法能够完成不同场景的稀疏光场稠密重建任务。
【Abstract】 The Light-field can record the characteristics of high-dimensional light data in the real environment to perceive the changes in the surrounding dynamic environment to play the light field’s role in environmental decision-making in an intelligent system.Light field imaging technology breaks through classical imaging models and digital camera limitations.The expression of light field all-optical function defines the relationship of light in multiple dimensions such as time domain,space domain,and spectrum.It then completes the tasks of intelligent processing,such as coupling perception and decoupling reconstruction.At the same time,it enables the light field to image multi-dimensional and large-scale dynamic scenes.Because of its unique advantages,light field imaging technology has been widely used in the areas of Refocus,virtual reality,life science,and national security.However,with the discrete sampling of high-dimensional data,light field imaging technology is facing the fundamental problem of obtaining high spatial resolution and high angle resolution light field simultaneously.How to carry out high-quality,dense reconstruction of sparse angle dimension light field data has become an urgent problem to be solved.Aiming at the problem that light field imaging technology cannot obtain high spatial resolution and high angular resolution light field simultaneously,this paper researches the optical flow field’s sparse light field density reconstruction algorithm.Starting from the basic principle of two-dimensional optical flow and based on the biplane model of the light field,this paper proposes a new mathematical model of a four-dimensional optical flow constraint equation,which is used to determine the position coordinates of the new viewing angle.Through interpolation calculation and image inversion,the intensity value of new coordinates is obtained point by point,and finally,a high-quality new perspective synthetic image is obtained.Experiments in several different scenarios verify that the proposed algorithm has strong adaptability.At the same time,the quantitative evaluation results show that the algorithm can complete the task of sparse light field dense reconstruction in different scenes.