最小化预测残差的图像序列压缩感知.pdfVIP

第 20 卷第9 期 光学精密工程 Vo l. 20 No.9 2012 年 9 月 Optics and PrecÎsion Engineering Sep.2012 文章编号 1004-924XC20 12)09-2095-07 最小化预测残差的图像序列压缩感知 石文轩\李捷 (武汉大学电子信息学院，湖北武汉 430072) 摘要:提出了一种最小化预测残差的图像序列压缩感知算法以实现高速相机输出图像的实时压缩。首先，在编码端仅使 用映射矩阵对原始输出图像进行压缩，将压缩得到的观测向量通过信道传输到解码端。接着，在解码端对相邻帧进行运 动估汁和运动补偿，得到一幅待重建图像的预测图像，利用压缩感知算法对原始图像和预测图像之间存在的预测残差图 像进行重建。最后，用迭代的方法优化预测残差图像的重建结果，直到连续两次的重建结果之差小于设定阔值，从而获 得重建的原始图像。采用 DALSA 公司的 CR-GENO-H6400 相机进行的实验表明，该算法可以实现 1 000 framej s 图像 的实时压缩，并且图像重建质量比独立地重建每张图像至少提高了 2~6 dB，有效地实现了对高速相机输出图像的实时 压缩与高质量重建。 关 键 词:压缩感知;实时压缩;图像重建z 预测残差 中图分类号:TP751 文献标识码:A doi: 10. 3788jOPE. 2095 Image sequence compressed sensing by minimizing prediction errors SHI Wen-xuan铃， LI Jie (School of Electronic Information , Wuhan University. Wuhan 430072 , China) 头 Corresponding author , E-mail:shiwx@163.com Abstract: An image sequence compressed sensing algorithm by minimizing prediction errors was pro- posed for high speed camera image compression in real-time. First , an original image was compressed only by a projection matrix on th巳 encoder side. The observed vector obtained by compressing was transferred to the decoder through a channel. Then , motion estimation and motion compensation were performed on correlated images on the decoder side , and a prediction image was generated in this way. Furthermore. the prediction error image which is the difference between original image and prediction image was reconstructed by compressed sensing. Finally , the reconstruction of prediction error image was improved by an iterative procedure , until the difference between t