摘要
为了改进Retinex算法在阴影消除中的不足,该文提出了一种基于高斯分布与Retinex模型的阴影检测与消除算法,通过灰度分布将图像分割为阴影区域、半影区域和正常光照区域,采用Retinex算法分别对阴影区域和正常光照区域提取光照,并消除光照影响,然后对半影区域进行双线性插值,得到消除阴影后的图像。该算法对大面积阴影的检测与消除具有很好的效果,同时避免了Retinex算法在灰度变化剧烈区域产生的模糊问题,实验证实了该算法的有效性。
In order to improve the ability of Retinex algorithm in shadow removal,a novel algorithm is proposed to detect and remove shadows based on Gaussian distribution and Retinex model.Using gray distribution information,original images are segmented to three regions：shadow,penumbrae and umbra regions.The Retinex algorithm is adopted in shadow and umbra regions separately to extract and eliminate the effect of the illuminant.After the bilinear interpolation is used for filling up the penumbra region,a shadow-free image is finally obtained.The proposed algorithm achieves good performances in shadow detection and removal,and the halo artifacts generated by the Retinex algorithm in the high dynamic range area is avoided.Experimental results demonstrate the effectiveness of this algorithm.
出处
《南京理工大学学报》
EI
CAS
CSCD
北大核心
2011年第1期1-5,共5页
Journal of Nanjing University of Science and Technology
基金
国家自然科学基金重大研究计划重点支持项目(90820306)
