基于微透镜阵列的光场图像深度估计

针对光场相机同时记录光线位置信息和角度信息,提出估计光场图像深度的新方法。根据光场摄像技术的数字重聚焦原理,产生场景不同焦深的图像序列;采用基于目标区域的可变窗口清晰度评价算子对目标在图像序列中的聚焦度进行测量,通过高斯插值获得最清晰的成像位置作为深度求取结果。以基于微透镜阵列的Lytro光场相机为例,实验结果表明,在降低现有算法时间复杂度的基础上,该方法提取的深度图可靠,证明了该方法的可行性与优越性。

Depth map estimation of micro lens array light field photography

Contrast with traditional digital camera, light field camera captures both the space and angle information of the ray of light traveling across the camera, puts forward a new method to estimate the depth of light field images. According to the principle of light field photography technology and digital refocusing, produces a series of images of the scene that focused at different depth; Adopts clarity evaluation operator based on the target area and variable window size to measure the focus of target in image sequence, then search for the most focus image position combined with Gaussian interpolation to calculate the depth of target. Showed the camera of Lytro, for example, besides reducing the existing algorithm time complexity, proves the feasibility and effectiveness of this method to gain accurate depth map of scenes.