Text detection in natural scene images becomes highly demanded for unstructured data in banking. In this paper, we propose a new deep learning algorithm called MSER, Hu-moment and Deep learning for Text detection (MHDT) based on Maximum Stable Extremal Regions (MSER) and Hu-moment features. Firstly, we extract MSERs as candidate characters. Secondly, a character classifier is introduced with Hu-moment features to reduce the number of input for clustering. After single linkage clustering, a text classifier trained from a Deep Brief Network is used to delete non-text. The proposed algorithm is evaluated on the ICDAR database, and the experimental results show that the proposed algorithm yields high precision and recall rate.
|Title of host publication||ACM International Conference Proceeding Series|
|Publisher||Association for Computing Machinery|
|Number of pages||6|
|Publication status||Published - 22 Feb 2019|
|Event||11th International Conference on Machine Learning and Computing, ICMLC 2019 - Zhuhai, China|
Duration: 22 Feb 2019 → 24 Feb 2019
|Name||ACM International Conference Proceeding Series|
|Conference||11th International Conference on Machine Learning and Computing, ICMLC 2019|
|Period||22/02/19 → 24/02/19|
Bibliographical noteFunding Information:
This work is partially funded by the Fujian Fumin Foundation and is supported by the National Natural Science Foundation of China under Grant (No. 61672170 and No. 61871313), the Science and Technology Planning Project of Guangdong Province (No. 2017A050501035), and Science and Technology Program of Guangzhou (No. 201807010058).
© 2019 Copyright is held by the owner/author(s). Publication rights licensed to ACM.
- Deep learning
- Text detection
- Unstructured data