Text Localization and Recognition of Chinese Characters in Natural Scenes Based on Improved Faster Region-Based Convolutional Neural Network

Abstract

To solve the problems faced by Chinese character recognition, such as complex shapes and diverse structures, this paper adopts the Visual Geometry Group 16 (VGG-16) model for feature extraction and introduces a two-layer bidirectional Long Short-Term Memory (LSTM) network. It improves the Faster Region-based Convolutional Neural Network (Faster R-CNN) by using a Region Proposal Network (RPN) to extract candidate boxes and adjust the positions of candidate regions. The improved model, namely Faster BLSTM-CNN, is tested through three types of experiments: validation of feature extraction effectiveness, comparative analysis of the algorithm before and after improvement, and comparison with traditional recognition algorithms. Finally, an experimental comparison combining text recognition and localization is conducted. On the ReCTS and MSRA-TD500 datasets, the proposed Faster BLSTM-CNN achieves excellent performance in Chinese character localization and recognition: in natural scenes (ReCTS dataset), its total image identification rate (TIIR) reaches 81.54%, recognition precision rate is 88.14%, and inference speed is 86 ms per image. Compared with the end-to-end benchmark algorithms DETR+CRNN-RMC and CTPN+CRNN-RMC, it improves the recognition precision rate by up to 7.54%, the TIIR by up to 10.04%, and the inference speed by up to 21%. These gains are mainly attributed to the two-layer bidirectional LSTM for contextual feature modeling and the element-wise addition fusion strategy that avoids dimension explosion.