基于小波变换和神经网络的数字图像清晰度识别算法

作者：沈国鑫,陈国金,费海波 日期：2008-04-23/span> 浏览：3244 查看PDF文档

基于小波变换和神经网络的数字图像清晰度识别算法
沈国鑫,陈国金,费海波
(杭州电子科技大学 机械工程学院，浙江 杭州 310037)

摘 要：数字图像的清晰度评价一直是各类数字成像系统的一个关键问题。在研究了常用的清晰度评价方法的基础上，通过研究找出了图像二维小波分解向量与图像质量的对应关系，分析了其细节信息，并采用统计的方法提取了图像特征二层小波分量的统计值作为分析特征，然后构建了3层BP神经网络模型对图像清晰度进行识别，得到图像质量等级。仿真实验结果表明，这种方法可以达到较好的图像识别率。
关键词：小波分析；清晰度；神经网络；图像识别

Digital image quality recognition algorithm based on neural network and wavelet transformation
SHEN Guoxin, CHEN Guojin, FEI Haibo
(College of Mechanical Engineering, Hangzhou Dianzi University, Hangzhou 310037, China)

Abstract: Evaluation of definition for gray scale digital image is an important aspect of digital imaging system. After the commonly sharpness function was analysised, the relations between twodimensional wavelet decomposition vector and the picture quality were studied and discovered, and the image characteristics from the detail information through statistics analysis were abstracted. Then 3 BP nerve network model for quality pattern recognition was costructed. The recognition rate of this means is good.
Key words: wavelet analysis; sharpness; neural network; image recognition
参考文献(Reference):
［1］罗发龙,李衍达.神经网络信号处理［M］.北京:电子工业出版社，1993.
［2］黄德双.神经网络模式识别系统理论［M］.北京:电子工业出版社，1996.
［3］张兆礼,赵春辉.现代图像处理技术及Matlab实现［M］.北京:人民邮电出版社,2001.
［4］焦李成.神经网络的应用与实现［M］.西安:西安电子科技大学出版社,1996.
［5］李 奇，冯华君，徐之海，等.数字图像清晰度评价函数研究［J］.光子学报，2002，31(6):736-738.
［6］LI Xi, LIU Guosui, NI Jinlin. Autofocusing of ISAR images based on entropy minimization［J］. IEEE Transactions on Aerospace and Electronic System,1999,35(4):1240-1252.
［7］GRANRATH D J. The role of human visual models in image processing［J］. IEEE〖WTBZ〗,1981,69(5):552-561.
［8］NILL N B. A visual model weighted cosine transform for image compression and quality assessment［J］. IEEE Trans. Commun,1985,33(6):551-557.