摘要
传统的重构矩阵优化算法可以事前确定稀疏变换基,却无法事前确定测量矩阵,尽管能够提高可压缩信号的重构效果,但不利于测量矩阵的硬件实现。针对这些问题,基于事前确定的稀疏变换基和单像素相机测量矩阵,提出了无需把二维影像转化成一维信号的基于线阵推扫和重构矩阵优化的单像素相机影像采集和重构方法。该方法实现了以事前确定的单像素相机测量矩阵采集数据,以事后优化的重构矩阵重构信号。初步解决了针对影像的测量矩阵硬件实现难题,改善提高了影像重构效果。同时回避了"测量矩阵和稀疏变换基不相关性好"的测量矩阵设计原则。实验表明无需把二维影像转化成一维信号的基于线阵推扫和重构矩阵优化的单像素相机影像采集和重构方法是可行的。
Traditional reconstruction matrix optimization algorithm can determine sparse transform base beforehand, but can not determine measurement matrix in advance. Although effect of the reconstructed compressible signal can be improved, but hardware design of measurement matrix is difficult. Based on pre-determined sparse transform base and single-pixel camera measurement matrix, an image acquisition and reconstruction method of single pixel camera based on line array push broom and matrix optimization reconstruction without changing 2D image into one-dimension signal was proposed to solve these problems. Collecting data by pre-determined singlepixel camera measurement matrix and reconstructing signal by optimized reconstruction matrix afterwards were implemented by this method. The problem of hardware design difficulty of measurement matrix for image was solved initially, and the effect of image reconstruction was improved. The method avoided measurement matrix design principle that the mutual-coherence of measurement matrix and sparse transform base must he good. Ex- periments results showed that the image acquisition and reconstruction method of single pixel camera based on line array push broom and matrix optimization reconstruction without changing 2D image into one-dimension signal was feasible.
出处
《探测与控制学报》
CSCD
北大核心
2014年第4期30-35,39,共7页
Journal of Detection & Control
基金
国家自然科学基金资助(41461082)
广西自然科学基金资助(2014GXNSFAA118285)
广西高校科学技术研究项目资助(YB2014212)
广西科技大学博士基金项目资助(校科博13212)
关键词
压缩感知
单像素相机
重构矩阵优化
影像采集
影像重构
离散余弦变换矩阵
compressive sensing
single-pixel camera
reconstruction matrix optimization
image acquisition
im age reconstruction
discrete cosine transform matrix