基于小波变换的模糊噪声图像的鉴别方法  

Blurred and noise images detecting based on wavelets transform

作　　者：岳冬雪[1] 徐婉莹[1] 黄新生[1] 

机构地区：[1]国防科学技术大学机电工程与自动化学院,长沙410073

出　　处：《计算机工程与应用》2007年第17期94-96,163,共4页Computer Engineering and Applications

摘　　要：根据Canny边缘检测算子的最优准则,模糊和噪声都会影响图像目标识别的精度。利用图像灰度的标准差定量地分析了模糊和噪声的影响;从频域角度分析了模糊和噪声的特性,并根据其特性提出了一种基于小波变换的模糊噪声图像的鉴别方法。该方法充分利用了小波的多分辨率分析特性和噪声的高频及模糊图像的低频特性,实验结果证明该方法能有效地鉴别图像的好坏。According to Canny＇s criteria of optimal edge detector,blurred image and noise-image reduce the recognizing precision.The property of blurred image and noise-image are quantitatively analyzed by the root-mean-square deviation of image gray level and qualitatively analyzed in frequency domain.Based on the property acquired above,a method of detecting blurred image and noise-image is proposed.The method made full use of the property of multi-resolution of wavelets to analyze the blurred image and noise-image.The results of the experiment indicate that the method proposed is reliable to discriminate the ＂good＂ images from ＂bad＂ ones.

关 键 词：图像质量 边缘检测 小波变换 

分 类 号：TN911.73[电子电信—通信与信息系统]