| Autor: |
Zhanghao Huang, Zhuo Su, Jichao Yan, Xiaonan Luo |
| Rok vydání: |
2016 |
| Předmět: |
|
| Zdroj: |
2016 6th International Conference on Digital Home (ICDH). |
| DOI: |
10.1109/icdh.2016.012 |
| Popis: |
Haze is a common atmospheric phenomenon in our dairy life. Image taken in foggy environment will have a loss of contrast and color due to the effect of haze. The hazed image would make a big trouble to some vision-driven applications since low accuracy in the scene recognition or object detection. Especially, single image dehazing is one of the challenging issue in the scene dehazing problem. In this paper, we propose a priorsmerging method for single image dehazing. We merge different haze-related priors in a learning framework and use it to achieve the haze-free image. There are two main stages in our method. The first stage is used to extract haze-related priors from haze image, the second aims to merge these priors through a learningbased framework. According to the experimental analysis, we demonstrate our method has more superior performance than the state-of-the-art dehazing methods. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
