Kanji Recognition in Scene Images without Detection of Text Fields - Robust Against Variation of Viewpoint, Contrast, and Background Texture.
Yoshinori KusachiAkira SuzukiNaoki ItoKenichi ArakawaPublished in: ICPR (1) (2004)
Keyphrases
- complex background
- scene images
- scene text
- partial occlusion
- viewpoint
- natural scene images
- text regions
- text extraction
- text detection
- lighting conditions
- text recognition
- text information
- text segmentation
- text lines
- feature extraction
- face detection
- illumination conditions
- optical character recognition
- color distribution
- license plate
- region detection
- bounding box
- connected components
- foreground objects
- computer vision
- character recognition
- natural scenes
- video frames
- object tracking
- mean shift
- object detection
- input image
- image segmentation
- automatic recognition
- object recognition
- image retrieval
- document images
- infrared images
- texture information