基于深度范例的中文文本识别方法
By employing a deep paradigm-based Chinese text recognition method, which utilizes paradigm queries and reordering networks, the problem of similar-looking characters in Chinese text recognition is solved, thereby improving recognition accuracy and reducing resource requirements, achieving efficient Chinese text recognition.
CN118247796BActive Publication Date: 2026-07-17FUDAN UNIVERSITY +1
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- FUDAN UNIVERSITY
- Filing Date
- 2024-02-20
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing technologies struggle to effectively identify similar-looking characters in Chinese text, resulting in low recognition accuracy. Furthermore, Chinese text recognition models have high demands on graphics card resources.
Method used
A deep paradigm-based Chinese text recognition method is adopted, including a paradigm query network and a paradigm re-ranking network. By combining feature extraction, global features and local features, and using a paradigm library for re-ranking, the recognition results are optimized.
Benefits of technology
It improves the accuracy and efficiency of Chinese text recognition, reduces resource consumption, and enhances the scalability of the model.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN118247796B_ABST
Abstract
本发明公开了一种基于深度范例的中文文本识别方法；该方法主要包含两个部分：范例查询阶段与范例重排序阶段。范例查询阶段主要用于预测识别结果并且定位每一个文字在图像上的位置，同时使用训练集组建成范例库，该范例库包含了每个文字的全局特征和局部特征；范例重排序阶段，主要针对范例查询阶段识别错误的情况，利用在范例库中检索与重排序得到更加准确的结果，纠正形近字的识别错误。本发明通过两个阶段的结合，成功提升了中文文本识别的准确性和效率，为中文识别领域的技术发展贡献了有力的解决方案。
Need to check novelty before this filing date? Find Prior Art