An Arabic handwriting synthesis system
作者：
Highlights：
• Expanding the training set for handwriting recognition can be beneficial even when the added data is synthesized from the original training data.
• Even simple shuffling-and-concatenation of characters can improve recognition accuracy.
• The usage of synthetic connection-strokes may increase the flexibility of the character-selection step.
• The recognition rate improved significantly due to our handwriting synthesis system.
摘要
•Expanding the training set for handwriting recognition can be beneficial even when the added data is synthesized from the original training data.•Even simple shuffling-and-concatenation of characters can improve recognition accuracy.•The usage of synthetic connection-strokes may increase the flexibility of the character-selection step.•The recognition rate improved significantly due to our handwriting synthesis system.
论文关键词：Arabic handwriting,Handwriting synthesis,Training data,Offline text recognition,Statistical model
论文评审过程：Received 3 May 2014, Revised 7 August 2014, Accepted 15 September 2014, Available online 28 September 2014.
论文官网地址：https://doi.org/10.1016/j.patcog.2014.09.013
