基于图像间相关性的光场压缩感知  

Light field compressive sensing based on the correlations across images

作　　者：牛涛[1] 沈为[1] 张之江[1] 雷蕊[1] 岳贤军[1] 

机构地区：[1]上海大学通信与信息工程学院,上海200072

出　　处：《电子测量技术》2014年第3期58-61,65,共5页Electronic Measurement Technology

摘　　要：介绍了一种基于光场图像间相关性的压缩感知方法。首先将光场图像按照相应拍摄相机的顺序堆叠在一起,形成一个图像集。然后沿着每一行对图像集进行切割,每个切割面会形成一幅切割图像。每个切割图像具有带状条纹的特征,且带状呈平行四边形。尽管带状条纹的大小不一样,但是它们的倾斜角保持一致。带状条纹内所有像素值近似相等,因此带状条纹内信号可以近似看成直流信号。将切割图像沿着带状条纹倾斜角进行重新排列,使每一列都可以近似看成直流信号,最后对重排序图像进行感知和重建,并重构原始图像。实验表明使用该方法重构图像的质量明显高于直接对原始图像进行压缩感知所重构图像的质量。This paper introduces a compressive sensing method based on the correlations across the light-field images.Firstly,stack all light-field images together to form an image volume ordered by the corresponding capturing camera.Then cut the image volume along each line,so every cutting surface forms a cutting image.Every cutting image has a characteristic with ribbon grain which is parallelogram.Although the size of ribbon grain is not the same,but their inclination angles remain the same.All the pixel values within ribbon grain approximately equal,so the signal within ribbon grain can be approximated as a direct-current（DC） signal.Rearrange the cutting image along the inclination angle of ribbon grain and make each column similar with DC signal.Finally measure the rearranged image and reconstruction it.Experiments show that the quality of the image reconstructed by using the method introduced in this paper is significantly higher than the quality of the image reconstructed by measureing the original image and reconstruction directly.

关 键 词：光场图像 压缩感知 图像间相关性 

分 类 号：TN911.74[电子电信—通信与信息系统]