结合图像结构特征和近似l_0范数的压缩采样恢复算法  

Compressive Sampling Image Recovery with Structure Features and Approximate l_0 Norm

作　　者：尚斐[1,2] 杜慧茜[3] 贾云得[2] 

机构地区：[1]北京理工大学生命科学学院,北京100081 [2]北京理工大学计算机学院智能信息技术北京市重点实验室,北京100081 [3]北京理工大学电子与信息学院,北京100081

出　　处：《计算机辅助设计与图形学学报》2010年第11期1874-1879,共6页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(90920009;60905006);国家"八六三"高技术研究发展计划(2009AA01Z323)

摘　　要：为了从压缩采样数据快速有效地恢复自然图像,提出了一种结合近似l0范数和近似总体变分(TV)的压缩采样图像恢复算法模型——TVSl0,并在恢复算法中引入模拟退火方法来实现快速恢复.该模型以最小化近似l0范数为基础,融入了反映图像结构特点的近似TV范数,体现出该模型对图像空域变化有限这一特点的适应性;并使用连续近似函数解决了l0范数的不连续问题.针对典型自然图像恢复的实验结果验证了文中算法的有效性和可行性,其恢复质量和基本TV模型的方法相当,但迭代次数少、计算复杂度低.This paper presents a new model named TVSl0 for natural image recovery from compressive samples.The model combines total variation norm and approximate l 0 norm.Simulated Annealing is employed to achieve optimization.The model is based on the approximate l 0 norm,in which the approximate function is used to tackle the discontinuity of l 0,and the approximate TV norm reflects the image structure features,i.e.bounded variation in space domain.The simulation results show that the natural images could be recovered rapidly and accurately.Comparing with TV minimization model,TVSl 0 can provide the recovery images in the same quality,with smaller number of iteration and lower complexity.

关 键 词：压缩采样 图像恢复 l0范数 TV范数 模拟退火 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]