科技通报
科技通報
과기통보
BULLETIN OF SCIENCE AND TECHNOLOGY
2014年
10期
133-135
,共3页
Gabor传递函数%相似性测度%图像质量%评价
Gabor傳遞函數%相似性測度%圖像質量%評價
Gabor전체함수%상사성측도%도상질량%평개
Gabor transfer function%similarity measure%image quality%evaluation
数字图像在压缩和融合等处理中会产生退化效应,当前对图像退化中质量评价指标如均方误差(MSE),峰值信噪比(PSNR)等不能有效反映图像在退化过程中传递性能,而只是单纯计算图像之间灰度值差异,对图像质量评价不准。提出一种利用Gabor传递函数相似性测度为评价指标的图像质量评价算法,采用图像的亮度、对比度和结构相似度为场景结构基准特征,设计Gabor传递函数。通过双正交滤波器得到Gabor传递函数相似性测度,该指标能有效反映图像退化演变过程中的传递性能,合理评价参考图像和失真图像质量。仿真实验表明,利用Gabor传递函数相似性测度指标评价图像质量,能准确反映图像失真等级,在几乎所有失真类型上都能有较满意的结果,评价结果比较稳定。
數字圖像在壓縮和融閤等處理中會產生退化效應,噹前對圖像退化中質量評價指標如均方誤差(MSE),峰值信譟比(PSNR)等不能有效反映圖像在退化過程中傳遞性能,而隻是單純計算圖像之間灰度值差異,對圖像質量評價不準。提齣一種利用Gabor傳遞函數相似性測度為評價指標的圖像質量評價算法,採用圖像的亮度、對比度和結構相似度為場景結構基準特徵,設計Gabor傳遞函數。通過雙正交濾波器得到Gabor傳遞函數相似性測度,該指標能有效反映圖像退化縯變過程中的傳遞性能,閤理評價參攷圖像和失真圖像質量。倣真實驗錶明,利用Gabor傳遞函數相似性測度指標評價圖像質量,能準確反映圖像失真等級,在幾乎所有失真類型上都能有較滿意的結果,評價結果比較穩定。
수자도상재압축화융합등처리중회산생퇴화효응,당전대도상퇴화중질량평개지표여균방오차(MSE),봉치신조비(PSNR)등불능유효반영도상재퇴화과정중전체성능,이지시단순계산도상지간회도치차이,대도상질량평개불준。제출일충이용Gabor전체함수상사성측도위평개지표적도상질량평개산법,채용도상적량도、대비도화결구상사도위장경결구기준특정,설계Gabor전체함수。통과쌍정교려파기득도Gabor전체함수상사성측도,해지표능유효반영도상퇴화연변과정중적전체성능,합리평개삼고도상화실진도상질량。방진실험표명,이용Gabor전체함수상사성측도지표평개도상질량,능준학반영도상실진등급,재궤호소유실진류형상도능유교만의적결과,평개결과비교은정。
Digital image compression and fusion process produces will result in degradation effect, the current image quality evaluation index such as the mean square error (MSE), peak signal to noise ratio (PSNR) and other objects cannot effective-ly reflect the performance degradation process of passing. The image quality evaluation is bad. An image quality evaluation algorithm called Gabor transfer function r similarity measure was proposed. The image brightness, contrast and structural similarity to the reference scenario structural characteristics were extracted. Gabor transfer function was designed. Similari-ty measure was got based on Gabor transfer function and biorthogonal filters. The index can effectively show the evolution of image degradation transfer performance evaluation of the reference image and the distortion reasonable image quality. Simulation results show that this algorithm can reflect the level of distortion on almost all types of distortions, and the evalu-ation results more stable.