视觉传感网中基于二次规划的图像压缩感知  

Image Compressed Sensing Based on Quadratic Programming in Visual Sensor Networks

作　　者：周钦青[1] 陈遵德[1] 

机构地区：[1]顺德职业技术学院电子与信息工程系,广东佛山528333

出　　处：《计算机工程》2014年第3期258-261,265,共5页Computer Engineering

基　　金：广东省信息技术教指委立项基金资助项目(XXJS-2013-35);顺德职业技术学院教改基金资助项目(2012-SZJGXM20)

摘　　要：为降低视觉传感网络中图像压缩感知算法的计算复杂度,提出一种基于二次规划的网络图像恢复算法。该算法将压缩感知重构中的欠定线性方程组求解问题,转化为有界约束二次规划问题,在此基础上结合阿米霍步长准则,设计一种压缩感知图像恢复算法,通过求解二次规划问题对网络图像数据进行恢复。理论分析和仿真结果表明,与传统图像压缩感知算法相比,该算法可减少约1/3的图像数据恢复运算时间,且图像重构质量提高3 dB^6 dB,有效提高了视觉传感器网络图像恢复算法的实时性。For reducing the computational complexity of Compressed Sensing（CS） in Visual Sensor Networks（VSN）, an image recovery algorithm based on quadratic programming is proposed. The under-determined linear equations in CS recovery are solved by bound-constrained quadratic programming, and an image recovery algorithm is designed based on the armijo rule. Theory analysis and experimental results show that the proposed algorithm can reduce 1/3 operation time of image recovery, and enhance the recovery quality by 3 dB-6 dB, thus significantly improve the real-time performance of image recovery in VSN.

关 键 词：视觉传感网 压缩感知 图像恢复 二次规划 有界约束 实时性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]