摘要
研究了多尺度Retinex算法对曝光量不足的彩色图像增强处理的结果。改进型多尺度Retinex算法是使用多尺度Retinex增强后在亮度平均值附近以k倍标准差进行截取、拉伸。采用亮度与对比度乘积、图像信息熵等2个判据,实验结果表明在亮度平均值附近k=1倍标准差进行截取后再拉伸得到的图像其2个判据的值都较大,图像最佳。
The results of the handled color images using multi-scale Retinex for the images that the time of exposure is not enough are studied.The advanced multi-scale Retinex algorithm was a process that it stretched the range of k times standard deviation near the brightness mean after the images handled by multi-scale Retinex.The results of experiment show the image are best image when the process is that it streteched the range of k=1 time standard deviation near the brightness mean because the two criterions are bigger.One of the two criterions is the criterion of production of brightness multipling with contrast degree,amother criterion is informatiom entropy.
出处
《光学技术》
CAS
CSCD
北大核心
2008年第S1期69-73,共5页
Optical Technique
基金
江苏省现代光学技术技术重点实验室开放课题资助项目(KJS0730)
关键词
信息光学
截取区间
多尺度RETINEX
信息熵
information optics
interception range
multiscale Retinex
information entropy