TextDCT: Arbitrary-Shaped Text Detection via Discrete Cosine Transform Mask.
Yuchen SuZhiwen ShaoYong ZhouFanrong MengHancheng ZhuBing LiuRui YaoPublished in: CoRR (2022)
Keyphrases
- discrete cosine transform
- text detection
- arbitrary shaped
- image compression
- dct coefficients
- transform domain
- object recognition
- arbitrary shape
- connected components
- text information
- video analysis
- filter bank
- urban scenes
- packing problem
- computer vision
- frequency domain
- gray scale
- video data
- wavelet transform
- feature space
- feature selection