基于Retinex的图像阴影恢复技术的研究与实现  

Research on image technology of shadowed region recovery based on multi-scale Retinex

作　　者：王潇潇[1] 孙永荣[1] 张翼[1] 刘晓俊[1] 

机构地区：[1]南京航空航天大学自动化学院导航研究中心,南京210016

出　　处：《计算机应用研究》2013年第12期3833-3835,共3页Application Research of Computers

基　　金：中央高校基本科研业务费专项资金资助项目(NP2011049)

摘　　要：针对阴影部分细节恢复的Retinex模型,提出了一种将多尺度Retinex算法与泰勒拉伸相结合的新型算法。算法中采用高斯滤波无限脉冲响应实现多尺度Retinex,并且通过一元二次泰勒展开函数对图像进行拉伸,同时利用高斯分布对拉伸区域进行自适应设定。实际结果表明,提出的算法能够有效地增强图像信息,在恢复和改善阴影区域细节的同时提高了图像的对比度,并保持了图像的色彩空间。同时,本算法具有较快的运算速率,在一定程度上实现了算法的实时性要求。Based on the Retinex model for shadowed region recovery, this paper proposed a new algorithm combined multiscale Retinex algorithm and Taylor-stretching. The algorithm adopted the infinite impulse response（IIR） of Gaussian filter to actualizes the umlti-scale Retinex model, extended the image by monte function with second power Taylor series and set the stretching interval adaptively based on Gaussian distribution. Experimental results show that the algorithm can enhance the image effectively, and recover the details of the shadowed region while increasing the image contrast and keeping the color space perfectly. Also, it has high operating rates which realize the real-time demand.

关 键 词：阴影恢复技术 多尺度RETINEX 泰勒拉伸 高斯滤波无限脉冲响应 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]