Image Haze Removal Using Computational Intelligence Methods
碩士 === 國立勤益科技大學 === 資訊工程系 === 102 === The image quality will be reduced with the light attenuation by atmospheric particles when taking a photograph outdoors, especially in the environment with haze. Because the hazy images lose a lot of information, the image recognition systems cannot recognize th...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2014
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/22352544333454728309 |
| Summary: | 碩士 === 國立勤益科技大學 === 資訊工程系 === 102 === The image quality will be reduced with the light attenuation by atmospheric particles when taking a photograph outdoors, especially in the environment with haze. Because the hazy images lose a lot of information, the image recognition systems cannot recognize the target in the image. In order to solve the problem of the image quality attenuation, there are two image dehazing methods proposed in this paper. To use the proposed fuzzy inference system is able to estimate the attenuation condition of the light. For the problem of the halo artifacts, this paper combines the morphology and the neural network to solve this problem. This paper proposes the average method to calculate the atmospheric light to solve the problem causing by the color cast. The other method is to use the fuzzy inference system to estimate the two different transmission maps which are combined by the weighted method to produce the recovered image. Finally, we demonstrate the experimental results and compare our proposed methods with other existing approaches.
|
|---|