论文部分内容阅读
基于大气散射的物理模型,针对雾霾的厚薄分布状况,给出了一种有选择性的雾霾去除新策略。在深入分析相关研究的基础上,新方法主要进行了两方面的改进,首先增加了抠图能量函数的数据项的精度,并增大了该能量函数的正则化参数值;其次对分布不均的雾霾影像使用自适应参数的云背景模型进行了预处理。改进方法的去霾结果更加清晰,色调、亮度分布均匀,而且色彩变化比较平滑自然,相比于已有同类算法,去霾效果更为理想。
Based on the physical model of atmospheric scattering, a new selective haze removal strategy is given according to the distribution of haze thickness. On the basis of further analysis of related researches, the new method is mainly improved in two aspects. Firstly, the accuracy of the data items of the energy function of matting is increased, and the regularization parameter value of the energy function is increased. Secondly, The haze image was preprocessed using a cloud background model with adaptive parameters. The haze removal result of the improved method is clearer, the color tone and brightness are evenly distributed, and the color change is relatively smooth and natural. Compared with the existing similar algorithms, the haze removal effect is more ideal.