DTRNet: Dual Text-Radical Decoding for Handwritten Chinese Text Recognition with Faked Character Detection

2026-08-06

AI Summary

This work addresses the dual challenge in K-12 educational settings of achieving both high-accuracy transcription and effective detection of forged characters in handwritten Chinese text—a task where existing methods struggle to balance efficiency with structural interpretability. To this end, the authors propose DTRNet, a novel architecture featuring a dual-branch decoding framework: a text branch performs line-level recognition, while a radical branch predicts Ideographic Description Sequences (IDS) that capture the semantic composition of Chinese characters. An IDS-Guided Confidence Adjustment (IGCA) mechanism then integrates structural evidence from the radical branch to refine recognition outputs. This approach uniquely unifies efficient line-level detection with radical-level interpretability, enabling robust forged character identification without compromising recognition performance, thereby overcoming the longstanding trade-off between efficiency and explainability in prior methods.

Abstract

In K-12 educational scenarios, handwritten Chinese text recognition should not only transcribe student writing, but also detect faked characters. However, existing recognition models are usually confined to a predefined set of normal characters and therefore cannot explicitly identify faked characters. Existing detection methods exhibit complementary limitations: character-level methods provide interpretable structural evidence but suffer from low efficiency, whereas line-level methods are efficient but rely heavily on confidence scores, making them prone to missed detections and lacking explicit structural evidence. Thus, the key challenge is to preserve character-structural evidence independent of contextual inference while maintaining line-level efficiency. To this end, we propose DTRNet, a dual Text-Radical decoding framework for line-level faked character detection. DTRNet decouples context-aware text recognition from character-wise structural verification, where the text branch performs line-level transcription and the radical branch predicts legal Ideographic Description Sequences (IDS) for lexicon-based faked character judgment. We further introduce IDS-Guided Confidence Adjustment (IGCA) to refine text predictions using structural evidence during inference. Experimental results demonstrate that DTRNet effectively detects faked characters while maintaining strong recognition performance and providing interpretable radical-level evidence. Code, checkpoints, and the processed dataset are publicly available at https://github.com/BNU-ERC-ITEA/DTRNet.

Problem

• Handwritten Chinese Text Recognition
• Faked Character Detection
• Character Structure
• Efficiency
• Interpretability

Innovation

• Dual Decoding
• Faked Character Detection
• Ideographic Description Sequence
• Handwritten Chinese Text Recognition
• Confidence Adjustment