2018 Fiscal Year Final Research Report
Accuracy improvement of the 3D rotation angle estimation of characters by the probabilistic interpolation
Project/Area Number |
16K00236
|
Research Category |
Grant-in-Aid for Scientific Research (C)
|
Allocation Type | Multi-year Fund |
Section | 一般 |
Research Field |
Perceptual information processing
|
Research Institution | Mie University |
Principal Investigator |
|
Co-Investigator(Kenkyū-buntansha) |
大山 航 九州大学, システム情報科学研究院, 准教授 (10324550)
|
Project Period (FY) |
2016-04-01 – 2019-03-31
|
Keywords | 文字認識 / 回転角度推定 / 車番認識 / カメラベース認識 / 確率的補間 |
Outline of Final Research Achievements |
We improved the precision of three-dimensional rotation angle estimation of the character without increasing memory usage by reducing feature vector size and employing the probabilistic interpolation of the rotational angle. And we proposed the camera-based license plate recognition for the image photographed from various direction by the matching between the combination of character candidates and the template of license plate. In addition, we proposed the method to detect and recognize kanji character consists of multiple connecting components by combinational optimization.
|
Free Research Field |
知覚情報処理
|
Academic Significance and Societal Importance of the Research Achievements |
記憶容量の削減と3次元回転角度推定の精度向上により、日本語の3次元回転不変文字認識が可能となり、従来は実現不可能であった日本語の各種看板や道路案内標識を様々な角度から自在に認識することができるようになる。その結果、動画からのリアルタイムの翻訳案内システムやカメラの設置場所を選ばない車番認識システムによる駐車場の入出管理など、カメラベース文字認識システムの発展に大きく寄与する。
|