WEKO3
アイテム
曇り度の自己調整によるヘイズ除去に関する研究
https://doi.org/10.15118/00005126
https://doi.org/10.15118/000051260d34641f-44a1-47b0-88d4-7a698605473e
| 名前 / ファイル | ライセンス | アクション |
|---|---|---|
A370 (3.5 MB)
|
|
|
|
A370_summary (353.1 kB)
|
|
| アイテムタイプ | 学位論文 / Thesis or Dissertation(1) | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 公開日 | 2015-06-11 | |||||||||||
| タイトル | ||||||||||||
| タイトル | Study on Image Dehazing with the Self-Adjustment of the Haze Degree | |||||||||||
| 言語 | en | |||||||||||
| タイトル | ||||||||||||
| タイトル | 曇り度の自己調整によるヘイズ除去に関する研究 | |||||||||||
| 言語 | ja | |||||||||||
| 言語 | ||||||||||||
| 言語 | eng | |||||||||||
| 資源タイプ | ||||||||||||
| 資源タイプ識別子 | http://purl.org/coar/resource_type/c_db06 | |||||||||||
| 資源タイプ | doctoral thesis | |||||||||||
| ID登録 | ||||||||||||
| ID登録 | 10.15118/00005126 | |||||||||||
| ID登録タイプ | JaLC | |||||||||||
| アクセス権 | ||||||||||||
| アクセス権 | open access | |||||||||||
| アクセス権URI | http://purl.org/coar/access_right/c_abf2 | |||||||||||
| 著者 |
毛, 鈞
× 毛, 鈞
|
|||||||||||
| 抄録 | ||||||||||||
| 内容記述タイプ | Abstract | |||||||||||
| 内容記述 | With the increase in industrial production and human activities, the concentration of atmospheric particulate matter is increasing substantially, leading to more frequent occurrence of fog and haze. Limited visibility caused by suspended particles in the air, such as fog and haze, is a major problem for many computer vision applications. The scenes captured by such computer vision systems suffer from poor visibility, low contrast, dimmed brightness, low luminance, and distorted color, which makes detection of objects within a scene more difficult. Therefore, visibility improvement, contrast, and feature enhancement of images and videos captured in inclement weather, which is referred to as dehazing, is inevitable. Haze removal is a difficult problem due to the inherent ambiguity between haze and the underlying scene. Model-based single-image dehazing methods are physically sound and produce qualitatively good results; however, more attention to dehazing quality rather than applicability leads to real-time applications that may not always be sufficiently fast. And in addition, the certain limitations exist, such as color bias and an inability of dealing with the sky area. The objectives of this study are to estimate the degree of haze from a single image automatically, and label the image with a haze factor, to propose a self-adjustment dehazing method using the degree of haze and evaluate the performance of the proposed dehazing method. In this research, based on atmospheric scattering model analysis and statistics of various outdoor images, the following conclusions can be drawn. Clear-day images have higher contrast than images plagued by bad weather. In addition, in most local regions, including the sky, hazed images have larger minimum values for most color channel pixels. Relying on these two observations, an estimate function that is related to a haze removal constant parameter has been developed to label a foggy image with different degrees of haze. By modifying the dark channel in advance, the proposed method reduces computational cost while providing promising dehazed results for real-time applications. The transmission estimation is performed by obtaining the minimum value only from a relevant pixel, or the mean filter of the minimum values of its neighboring pixels. The main advantage of the proposed algorithm compared to others is its processing speed. Its complexity is a linear function of the number of image pixels. Another advantage is similar image processing quality using the Dark Channel Prior method. Furthermore, labeling images with different haze-degree can make batch processing of an image set with hazy and haze-free images possible. Finally, the degree of haze removal can be adjusted adaptively with a built-in function that makes the dehazed images look more natural. The experimental results clearly indicate that the proposed approach achieves good performance for enhanced visibility, processing speed, and stability, which makes the proposed method applicable for real-time requirements. | |||||||||||
| 言語 | en | |||||||||||
| 学位授与機関 | ||||||||||||
| 学位授与機関識別子Scheme | kakenhi | |||||||||||
| 学位授与機関識別子 | 10103 | |||||||||||
| 学位授与機関名 | 室蘭工業大学 | |||||||||||
| 言語 | ja | |||||||||||
| 学位授与機関名 | Muroran Institute of Technology | |||||||||||
| 言語 | en | |||||||||||
| 学位名 | ||||||||||||
| 学位名 | 博士（工学） | |||||||||||
| 言語 | ja | |||||||||||
| 学位の種別 | ||||||||||||
| 言語 | ja | |||||||||||
| 値 | 課程博士 | |||||||||||
| 学位授与番号 | ||||||||||||
| 学位授与番号 | 甲第370号 | |||||||||||
| 報告番号 | ||||||||||||
| 言語 | ja | |||||||||||
| 値 | 甲第370号 | |||||||||||
| 学位記番号 | ||||||||||||
| 言語 | ja | |||||||||||
| 値 | 博甲第370号 | |||||||||||
| 学位授与年月日 | ||||||||||||
| 学位授与年月日 | 2015-03-23 | |||||||||||
| 日本十進分類法 | ||||||||||||
| 主題Scheme | NDC | |||||||||||
| 主題 | 007 | |||||||||||
| 著者版フラグ | ||||||||||||
| 出版タイプ | VoR | |||||||||||
| 出版タイプResource | http://purl.org/coar/version/c_970fb48d4fbd8a85 | |||||||||||
| フォーマット | ||||||||||||
| 内容記述タイプ | Other | |||||||||||
| 内容記述 | application/pdf | |||||||||||
