基于梯度域的光场全聚焦图像生成方法

A Method for Generating All-in-Focus Image of Light Field Based on Gradient Domain

光场相机存在着空间分辨率和角度分辨率相互制约、成像分辨率不高的问题.针对这一问题,将传统图像融合方法与光场重聚焦技术相结合,提出了一种基于梯度域的光场全聚焦图像生成方法.以Lytro相机光场数据为基础,通过光场重聚焦提取焦点堆栈图像,然后将焦点堆栈图像按亮度和色度通道分别进行处理:亮度通道融合采用基于小波分解的梯度积分算法,色度通道融合采用基于像素的色度通道加权求和方法.实验表明,该方法融合亮度通道执行效率高,融合色度通道简单有效,能够生成清晰的光场全聚焦图像,提高了光场图像空间分辨率.

The trade-off between spatial resolution and angular resolution is one of the main problems of the light field camera.To address this problem,combining the traditional image fusion method with the light field refocusing technology,we present a novel method for generating a light field all-in-focus image based on the gradient domain.Based on the light field data of the Lytro camera,the focal stack image is extracted through light field refocusing,and then the focal stack image is processed according to the luminance and chrominance channels.A gradient integration algorithm based on wavelet decomposition is used to achieve luminance channel fusion and a pixel-based channel weighted summation method is used to realize the chrominance channel fusion.Experimental results show that the method proposed here has high efficiency in the fusion of the luminance channel,and the fusion of the chrominance channel is simple and effectiveand that it can generate a clear all-in-focus image of the light field and improve the resolution of the light field image.