Detection of Characters and their Boundary from Images of Modern Japanese Official Documents Using Fully CNN-based Filter
Kei Watanabe, Shinji Takahashi, Yuhei Takagi, Masashi Yamada, Yoshito Mekada, Junichi Hasegawa, Takatoshi Naka, Shinya Miyazaki · 2018
An automatic decipherment system for modern Japanese official documents has been developed in this research project in which a method for detecting characters and their boundary using a Fully CNN-based filter is initiated. The accuracy of detecting a boundary was observed to be more than 97% in our experiment.