摘要
利用光的偏振特性去除雾天图像雾霾是图像复原的一种重要方法 .本文在传统偏振图像复原原理的基础上,提出一种基于暗原色原理改进的偏振图像复原方法 .该方法根据雾天图像直方图呈近似正态分布,采用相应算法分离图像天空区域,估算大气光数值;通过改进的偏振图像暗原色先验方法,估算并优化透射率,并使用直方图均衡化方法调整算法参数,获取最优偏振图像透射率,最终根据去雾方程获取偏振图像复原结果 .实验测试结果及主客观评价表明,本文方法能够有效复原雾天图像信息,去雾效果较好.
It is an important method to remove haze from fog image by using the polarization of light. This paper proposes a new method to restore degraded polarization image using improved dark channel prior based on the principle of the traditional polarization image restoration. Based on the approximately normal distribution of the fog image histogram,we separated the image sky area to estimate the atmospheric light value with the corresponding algorithm. The transmittance was estimated and optimized by improved dark channel prior,and the histogram equalization method was used to adjust it in order to obtain the optimal polarization image transmission. The original clear image was obtained by the polarization image restoration model. The experimental results were also carried out in subjective visual evaluation and objective parameter evaluation. The results showed that this method could effectively improve the clarity of the image of haze.
出处
《绵阳师范学院学报》
2017年第8期49-56,共8页
Journal of Mianyang Teachers' College
基金
福建省自然科学基金项目(2016J01751)
福建省教育厅中青年教师教育科研科技类基金项目(JAT160693)
集美大学诚毅学院教育教学改革项目(C16067)
关键词
偏振光
正态分布
暗原色先验改进
透射率
直方图均衡化
Polarized light
normal distribution
improved dark channel prior
transmittance
histogram equalization