WEKO3
アイテム
極低照度状況と高密度散乱媒体状況下での3D画像可視化技術
https://doi.org/10.18997/00009164
https://doi.org/10.18997/0000916405662c27-072e-446a-bfca-3d715905c9b6
| 名前 / ファイル | ライセンス | アクション |
|---|---|---|
jou_k_376.pdf (96.0 MB)
|
|
| アイテムタイプ | 学位論文 = Thesis or Dissertation(1) | |||||||
|---|---|---|---|---|---|---|---|---|
| 公開日 | 2023-04-04 | |||||||
| 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 | |||||||
| タイトル | ||||||||
| タイトル | Three-dimensional (3D) image visualization techniques under the extremely dark situation and dense scattering media situation | |||||||
| 言語 | en | |||||||
| タイトル | ||||||||
| タイトル | 極低照度状況と高密度散乱媒体状況下での3D画像可視化技術 | |||||||
| 言語 | ja | |||||||
| 言語 | ||||||||
| 言語 | eng | |||||||
| 著者 |
Lee, Jaehoon
× Lee, Jaehoon
|
|||||||
| 抄録 | ||||||||
| 内容記述タイプ | Abstract | |||||||
| 内容記述 | Recently, three-dimensional （3D） image visualization has become an essential topic in image processing researches. Significantly, the 3D depth information acquisition and visualization under the low light condition or dense scattering media situation are needed in various industrial fields such as unmanned cameras and autonomous vehicles. To visualize the object under harsh conditions, radar and LiDAR have been used. LiDAR can visualize the object shape with discrete point clouds under a dark situation. However, LiDAR may not provide accurate 3D depth information well under inclement weather conditions, such as dense scattering media conditions. To compensate for these shortcomings, it is used with radar in various industries. Radar calculates the depth information using electromagnetic waves, but it only provides the grayscale depth map of the 3D object without object color information. Therefore, it cannot produce sufficient information to recognize the object accurately. On the other hand, the conventional camera can visualize the object with full color and calculate the 3D depth information using integral imaging. Integral imaging is a passive 3D reconstruction method that utilizes elemental images, which have different perspectives of 3D object. Moreover, photon counting has been used to acquire 3D object information under harsh conditions. It can measure the 3D object photons under a dark situation using a photon detector with high sensitivity performance. In addition, it can be used as the feature extractor to visualize the object under the deep scattering media condition. It also can be used for an optical encryption algorithm to keep the data securely. However, photon counting integral imaging has critical issues such as poor visualization performance by significantly sparing photon information and low 3D image visual quality, as explained below. First, photon counting integral imaging reconstruction process may not provide sufficient information in the severely dark situation. It utilizes the maximum likelihood estimation to visualize the object but can reduce the photon intensities. Therefore, it can cause information loss by reconstructing the 3D scene according to the depth. Second, photon counting image contains the measurement error and photon intensity fluctuations through the detector. The photon detector should have a high sensitivity sensor to detect the photons under the low luminance situation. High sensitivity performance is appropriate to detect the photons in the dark situation but can also generate errors and intensity fluctuations.These noises may decrease the 3D image quality under the dark and dense scattering media conditions. To solve these problems, we propose a novel photon counting integral imaging reconstruction method which can visualize the 3D scene even under the severely low-light situation. In addition, we figure out the noise photons and measurement error removal technique by analyzing the photon signal characteristic through the wavelet transform. Chapter 1 introduces the research background and needs for 3D visualization under photon-starved or deep scattering media conditions. We represent various methods to visualize the 3D scene and points out the related problems. Then, we briefly propose our solutions to enhance the 3D image visual quality. Chapter 2 briefly reviews integral imaging and photon counting technique utilized in this thesis. We present the basis of integral imaging technique and photon counting using equations and figures in detail. Then, we highlight the critical issues of photon counting integral imaging technique. Chapter 3 introduces a new reconstruction method that can effectively visualize 3D objects using the small number of photons and apply it to optical encryption, which can encrypt data more securely than conventional methods. We implement optical experiments to demonstrate our proposed method’s superiority and evaluate the image quality using image quality metrics. Chapter 4 presents the noise photons and error removal technique by using wavelet transform. We explain the photon characteristics and show the photon signal analysis result using wavelet transform. Then, we utilize it in photon counting integral imaging and scattering media removal techniques. To prove our noise removal technique and applications, we implement optical experiments and evaluate image quality metrics. Finally, we summarizes our research and presents the achieved objectives through research briefly and suggest the future research plan and objectives in chapter 5. | |||||||
| 目次 | ||||||||
| 内容記述タイプ | TableOfContents | |||||||
| 内容記述 | 1 Introduction||2 Conventional 3D visualization under photon-starved and dense scattering media conditions||3 Enhancing 3D image visual quality through the proposed reconstruction method||4 Enhancing 3D Image Visual Quality by Photon Signal Analysis||5 Conclusions | |||||||
| 備考 | ||||||||
| 内容記述タイプ | Other | |||||||
| 内容記述 | 九州工業大学博士学位論文 学位記番号: 情工博甲第376号 学位授与年月日: 令和5年3月24日 | |||||||
| キーワード | ||||||||
| 主題Scheme | Other | |||||||
| 主題 | 3D visualization | |||||||
| キーワード | ||||||||
| 主題Scheme | Other | |||||||
| 主題 | Photon counting | |||||||
| キーワード | ||||||||
| 主題Scheme | Other | |||||||
| 主題 | Scatter media removal | |||||||
| キーワード | ||||||||
| 主題Scheme | Other | |||||||
| 主題 | Integral imaging | |||||||
| キーワード | ||||||||
| 主題Scheme | Other | |||||||
| 主題 | Machine learning | |||||||
| アドバイザー | ||||||||
| 李, 旻哲 | ||||||||
| 学位授与番号 | ||||||||
| 学位授与番号 | 甲第376号 | |||||||
| 学位名 | ||||||||
| 学位名 | 博士（情報工学） | |||||||
| 学位授与年月日 | ||||||||
| 学位授与年月日 | 2023-03-24 | |||||||
| 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 | 17104 | |||||||
| 学位授与機関名 | 九州工業大学 | |||||||
| 学位授与年度 | ||||||||
| 内容記述タイプ | Other | |||||||
| 内容記述 | 令和4年度 | |||||||
| 出版タイプ | ||||||||
| 出版タイプ | VoR | |||||||
| 出版タイプResource | http://purl.org/coar/version/c_970fb48d4fbd8a85 | |||||||
| 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 | |||||||
| ID登録 | ||||||||
| ID登録 | 10.18997/00009164 | |||||||
| ID登録タイプ | JaLC | |||||||
