摘要
图像复原技术具有广泛的应用领域,近年来已经成为国内外图像界研究的热点问题之一。对于运动模糊图像的高质量复原更是在安保监控,交通监控等领域中急需解决的问题
本文主要研究了逆滤波、维纳滤波和Richardson-Lucy三种复原方法的原理,在有噪声(高斯、椒盐)和无噪声两种条件下分别对其做了匀速直线运动模糊恢复实验,并给出了图像质量评价的一些常用指标。实验结果表明:当图像中无噪声时,逆滤波复原的图像质量最高,其次为维纳滤波,RL方法复原的图像质量最低,但图像的清晰度仍然可以接受。在此试验结果的基础上,使用VC++设计开发了运动模糊图像复原系统并设计了易于实现的DSP板卡结构。
该系统具有易于实现,运行稳定等优点,解决了目前同类系统不能使用多类算法对运动模糊图像进行复原的问题,具有一定的创新性,在同类产品中处于领先地位。该产品具有良好的商业化前景,如果投入到目前市场中,预期会取得较好的社会效益和一定的经济效益。
运动模糊图像复原系统所涉及的知识面比较广,本文只对其中的三种关键算法作了讨论,并使用DSP硬件板卡加以算法的实现。以后将在系统功能的扩展、算法优化、DSP板卡设计等方面做进一步研究。
The image restoration technology has the widespread application domain, in recent years already became one of domestic and foreign image research hot topics. Regarding the movement hazy picture's high grade restoration is in the security monitoring, domains and so on transportation monitoring is urgently needed the solution the question.
This article has mainly studied the liftering, the Weina filter and the Richardson-Lucy three kind of restoration method principle, in had the noise (Gauss, spiced salt) and does not have under the noise two kind of conditions to make the uniform rectilinear translation to resume the experiment fuzzily separately, and has given picture quality appraisal some commonly used targets. The experimental result indicated: When in the image does not have the noise, the liftering restoration's picture quality is highest, next is the Weina filter, the RL method restoration picture quality is lowest, but the image clarity still might accept. In this test result's foundation, used the VC++ designs to develop the movement hazy picture restoration system and to design the DSP board card structure which easy to realize.
This system has easy to realize, movement stable and so on merits, solved the present similar system not to be able to use many kind of algorithms to carry on the restoration to the movement hazy picture the question, had certain innovation, was in the leading position in the similar products. This product has the good commercialization prospect, if invests into the present market, anticipated will obtain the good social efficiency and certain economic efficiency.
The movement hazy picture restoration system involves the aspect of knowledge is quite broad, this article has only made the discussion to three essential algorithms, and uses the DSP hardware board card to perform the algorithm realization. Later in system function's expansion, the algorithm will optimize, aspects and so on DSP board card design to do further studies.
引文
[1]赵荣椿.数字图像处理导论[M].西安:西北工业大学出版社,1996.
[2]张远鹏.计算机图像处理技术基础[M].北京:北京大学出版社,1996.
[3]明文华.运动模糊图像复原算法研究[D].硕士学位论文,安徽大学,2004年5月.
[4]刘微.运动模糊图像恢复算法的研究与实现[D].博士学位论文,中国科学院研究生院,2005年12月.
[5]马明.运动模糊图像的判定与恢复[D].硕士学位论文,大连理工大学,2006年12月.
[6]谢义方.数字图像复原算法研究[D].硕士学位论文,湖南大学,2004年3月.
[7]张云霞.运动模糊图像的复原与重构[D].硕士学位论文,大连理工大学,2006年12月.
[8]张媛.运动模糊图像的处理[D].硕士学位论文,安徽大学,2005年5月.
[9]李红阳.运动图像恢复[D].硕士学位论文,清华大学,2002年6月.
[10]周荣彪.门牌号的运动模糊图像复原及识别方法研究[D].硕士学位论文,天津大学,2006年1月.
[11]郑捷文,蔡利栋.匀速直线运动模糊图像的恢复[J].暨南大学学报,2002,23(3):42-45.
[12]尹建新,计时鸣,张利.相对运动所引起的模糊图像的模糊复原算法研究[J].浙江工业大学学报,2006,34(1):78-81.
[13]易小波,李浪.图像运动模糊复原的研究[J].衡阳师范学院学报,2005,26(3):88-90.
[14]高梅,陈树越.匀速直线运动模糊图像的恢复[J].计算机工程与应用,2004,31:59-60.
[15]李树涛,王耀南.图像椒盐噪声的非线性自适应滤波[J].中国图象图形学报,2000,5(12):999-1001.
[16]明文华,孔晓东,屈磊,梁栋.运动模糊图像的恢复方法研究[J].计算机工程,2004,30(7):133-135.
[17]贺卫国,黎绍发.匀速直线运动模糊长度的精确估计[J].计算机应用,2005,25(6): 1316-1320.
[18]王晓红,陈秀万,谭仲军,赵荣椿.一种有效的运动模糊图像恢复算法[J].计算机工程,2003,29(17):13-116.
[19]贾永红.计算机图像处理与分析[M].武汉:武汉大学出版社,2001.9.
[20]张媛,张铃,张燕平.一种运动模糊图像的复原法[J].微机发展,2005,15(6):28-32.
[21]马彪.运动图像模糊度参数的估计[J].辽宁大学学报,2005,32(4):376-378.
[22]陆原,胡楠.一种基于VC++的运动图像复原技术[J].山西电子技术,2006,01:39-40.
[23]JAMES G. NAGY, DIANNE P. O'LEARY. IMAGE RESTORATION THROUGH SUBIMAGES AND CONFIDENCE IMAGES[M].Kent State University,2002.
[24]Clement Shimizu, Amit Shesh, Baoquan Chen. Hardware Accelerated Motion Blur Generation[M].University of Minnesota Computer Science Department Technical Report,2003.
[25]Mohsen Ebrahimi Moghaddam, Mansour Jamzad. Linear Motion Blur Parameter Estimation in Noisy Images Using Fuzzy Sets and Power Spectrum[M].Department of Computer Engineering, Sharif Univer-sity of Technology,2006.
[26]Fu, K. S. Syntactic Pattern Recognition and Applications, Prentice Hall, Upper Saddle River, NJ,2001
[27]Nevatia, R. Machine Perception, Prentice Hall, Upper Saddle River, NJ,2005
[28]Pavlidis, T. Algorithms for Graphics and Image Processing, Computer Science Press, Rockville, MD,1999
[29]Rosenfeld, R. and Kak, A. C. [1982]. Digital Picture Processing,2nd ed., vols.1 & 2, Academic Press, NY.
[30]Andrews, H. C. and Hunt, B. R. Digital Image Restoration, Prentice Hall, Upper Saddle River, NJ,2007
[34]Pavlidis, T. Structural Pattern Recognition, Springer-Verlag, NY,2007.