摘要
解决明暗突变多区域光晕现象,对于提升降质图像质量具有重要意义。针对传统的光晕消除方法在处理高动态范围图像时,局部细节出现过增强,易在明暗对比强烈的部分产生大气光误判、光晕伪影以及运算耗时等问题,提出一种基于多尺度Retinex的明暗突变图像多区域光晕自适应消除方法。利用大气光校验方法对明暗突变图像候选大气光的有效性进行判断,去除高光区域的干扰,利用基于块偏移的透射率计算方法获得突变图像边缘保持的透射率,抑制光晕像素的数量。将明暗突变图像局部方差和复杂度构造引导滤波的自适应平滑增益,估计图像照度分量,在对数域对多尺度Retinex数学模型进行求解,获取消除图像多区域光晕和细节保持的多尺度反射分量。对反射分量根据灰度等级进行自适应增强,获得最终增强后的图像。实验结果表明,所提方法能够有效克服光照不均并自适应消除了光晕现象,运算速度远优于当前同类方法。
A method to adaptively eliminate multi -region light halo in light and shade mutation image based on multi - scale Retinex is presented. Firstly, the atmospheric light calibration method was used to judge the effective- ness of candidate atmospheric light in light and shade image, so as to remove the interference of highlight area. Then, the transmissivity calculation method based on block offset was used to obtain the transmissivity of the edge preservation of mutation image and suppress the number of light halo pixels. Moreover, the adaptive smoothing gain of guided filter was constructed based on the local variance and complexity of light and shade image and the illumina- tion component of image was estimated. Meanwhile, the multi - scale Retinex mathematical model was solved in the logarithmic domain to obtain multi - scale reflections component which eliminated the multi - region light halo and preserved detail of image. Finally, the reflection component was adaptively enhanced based on the gray level to obtain the final image. Simulation results show that the proposed method can effectively overcome the uneven illumination and adaptively eliminate the light halo phenomenon. Meanwhile, the computation speed is better than the current method.
作者
侯培文
HOU Pei - wen(Taiyuan University,Taiyuan Shanxi 030006,Chin)
出处
《计算机仿真》
北大核心
2018年第8期339-342,456,共5页
Computer Simulation
关键词
明暗
突变图像
多区域
光晕
自适应消除
Light and shade
Mutation image
Multi -region
Light halo
Adaptive elimination
