An Application of Text Detection and Recognition for Electronic Design Automation
Chih-Chang Yu, Hsiao‐Wei Chen, Tzu‐Ying Chen, Po‐Hao Chen, Hsu-Yung Cheng · 2020
This work proposed an application of text detection and character recognition which is used for improving electronic design automation (EDA). Images used in this work contain colored regions and texts enclosed by an external boundary such as a circle or a polygon. Instead of using complicated methods, this work adopts several efficient image processing methods to extract true text regions followed by optical text recognition (OCR). Since the desired texts are not in normal form, we proposed some mechanisms to classify the characters into several types: normal, superscript, and subscript. In the experiments, the accuracy of judging the location information of block text can reach 93.36%. The proposed design can significantly improve the application of text detection and recognition in EDA.