检索结果-内蒙古大学图书馆

scene text recognition That Eliminates Background and Character Noise Interference

APPLIED SCIENCES-BASEL 2025年第7期15卷 3545-3545页

作者： Tang, Shancheng Cao, Yaoqian Liang, Shaojun Jin, Zicheng Lai, Kun Xian Univ Sci & Technol Coll Commun & Informat Engn Xian 710054 Peoples R China

In natural photographs, complex background noise and character noise frequently interfere with scene text identification. To solve the aforementioned concerns, this paper proposes a novel scene character identification model that eliminates noise from both the backdrop and the character (ENBC). The model is divided into three pieces. To begin, the high-level character feature extraction module uses ASPP dilated convolution with varying expansion rates to obtain features at various scales, thereby expanding the receptive field to capture the character feature area more effectively, eliminating noise interference from the character itself, and improving the character shape features. Second, the multi-level character feature fusion module merged the high-level character feature information after upsampling with the low-level character feature information in the backbone network, separated the foreground characters from background interference, removed background noise, and output the resulting image. Third, the recognition enhancement module enhances character context modeling by considering both forward (left-to-right) and backward (right-to-left) information from the text sequence. The experimental results show that the model can effectively minimize background and character noise interference, boosting recognition accuracy by at least 4.2% on the synthetic scene dataset. When compared to other popular techniques on the IIIT5K, ICDAR-2015, ICDAR-2003, and CUTE80 public datasets, recognition accuracy improves by an average of 6.97%.

关键词： scene text recognition deep learning background noise character's own noise enhanced character features

来源：评论

学校读者我要写书评

暂无评论

scene text recognition Based on Corner Point and Attention Mechanism 21st

Scene Text Recognition Based on Corner Point and Attention M...

引用

21st Pacific Rim International Conference on Artificial Intelligence

作者： Wang, Hui Hu, Tao Geng, Xiaoke Li, Kai Huazhong Univ Sci & Technol Sch Comp Sci & Technol Wuhan 430074 Peoples R China

ISBN: (纸本)9789819601158;9789819601165

The irregular distribution of text is a significant challenge for current scene text recognizers. Rectification methods based on Thin-Plate Spline (TPS) are adopted by various recognizers due to their plug-and-play nature. However, since the TPS transformation parameters are determined based on text boundaries regressed, it struggles to focus on the position of text areas relative to the image and fails to suppress background interference. Although the rectified images retain the geometric structure of the text, this leads to unnatural character deformation and characters exceeding the image boundaries. To address these issues, this paper proposes a scene text recognition algorithm based on corner points and attention mechanisms. Firstly, a novel irregular text rectification algorithm named CornerTPS is introduced, which integrates corner features with image features, allowing the TPS transformation to focus on character position information and improving the rectification effect for irregular text. Then, CornerTPS is improved to a feature-level rectification, and the rectified corner feature map is integrated into the encoder, further guiding the encoder to focus on text area features and reducing the interference of non-text area corners. Experiments on public datasets demonstrate that the algorithm proposed in this paper has a significant advantage in irregular text recognition.

关键词： scene text recognition text rectification Thin-Plate Spline Corner point map Attention mechanism

来源：评论

学校读者我要写书评

暂无评论

scene text recognition Via k-NN Attention-Based Decoder and Margin-Based Softmax Loss 7th

Scene Text Recognition Via k-NN Attention-Based Decoder and ...

引用

7th Chinese Conference on Pattern recognition and Computer Vision

作者： Zhang, Hongxia Xu, Minqiang He, Liang Xinjiang Univ Sch Comp Sci & Technol Urumqi 830017 Peoples R China iFLYTEK Co Ltd Hefei 230088 Peoples R China Tsinghua Univ Dept Elect Engn Beijing 100084 Peoples R China

ISBN: (纸本)9789819785100;9789819785117

Facing various complex background and diverse text image shape, this paper proposes an encoder-decoder-based scene text recognition model named E2D-Rec to enhance the recognition capability of irregular text and achieve stronger generalization. Firstly, a text rectification network is introduced to transform irregular texts such as curved and skewed texts into relatively regular ones, which are iteratively learned as several control points of text regions within the model. Through control points to drive TPS interpolation transformation, rectified text images are obtained. Then, a modeling network based on the encoder-decoder architecture sequentially predicts text sequences in an auto-regressive manner. The visual encoder is utilized to generate image patch embedding from text image, and the visual-textual decoder learns the correlation between word embedding and image patch embedding via k-NN attention selection. Finally, during the training phase, a loss function based on inter-class penalty is adopted as the model's objective function. By widening the boundaries between different classes in the final label space mapping layer, the model learns deep features with high discriminative power. Experimental results validate the proposed model's improved recognition performance on the Union14M-Benchmark and six commonly used datasets.

关键词： scene text recognition Transformer k-NN Attention

来源：评论

学校读者我要写书评

暂无评论

Attention Guidance by Cross-Domain Supervision Signals for scene text recognition

引用

IEEE TRANSACTIONS ON IMAGE PROCESSING 2025年 34卷 717-728页

作者： Xue, Fanfu Sun, Jiande Xue, Yaqi Wu, Qiang Zhu, Lei Chang, Xiaojun Cheung, Sen-Ching Shandong Normal Univ Sch Informat Sci & Engn Jinan 250014 Peoples R China Shandong Univ Tradit Chinese Med Coll Intelligence & Informat Engn Jinan 250355 Peoples R China Shandong Univ Sch Informat Sci & Engn Qingdao 266237 Peoples R China Tongji Univ Sch Elect & Informat Engn Shanghai 200070 Peoples R China Univ Technol Sydney Fac Engn & InformationTechnol Sydney NSW 2007 Australia Univ Kentucky Dept Elect & Comp Engn Lexington KY 40506 USA

Despite recent advances, scene text recognition remains a challenging problem due to the significant variability, irregularity and distortion in text appearance and localization. Attention-based methods have become the mainstream due to their superior vocabulary learning and observation ability. Nonetheless, they are susceptible to attention drift which can lead to word recognition errors. Most works focus on correcting attention drift in decoding but completely ignore the error accumulated during the encoding process. In this paper, we propose a novel scheme, called the Attention Guidance by Cross-Domain Supervision Signals for scene text recognition (ACDS-STR), which can mitigate the attention drift at the feature encoding stage. At the heart of the proposed scheme is the cross-domain attention guidance and feature encoding fusion module (CAFM) that uses the core areas of characters to recursively guide attention to learn in the encoding process. With precise attention information sourced from CAFM, we propose a non-attention-based adaptive transformation decoder (ATD) to guarantee decoding performance and improve decoding speed. In the training stage, we fuse manual guidance and subjective learning to learn the core areas of characters, which notably augments the recognition performance of the model. Experiments are conducted on public benchmarks and show the state-of-the-art performance. The source will be available at https://***/xuefanfu/ACDS-STR.

关键词： Feature extraction text recognition Decoding Encoding Image segmentation Image coding Transformers Iron Electronic mail Computer vision scene text recognition attention guidance attention drift vision transformer feature fusion

来源：评论

学校读者我要写书评

暂无评论

SNFR: salient neighbor decoding and text feature refining for scene text recognition

引用

MACHINE VISION AND APPLICATIONS 2025年第2期36卷 1-13页

作者： Lu, Tongwei Fan, Huageng Chen, Yuqian Shao, Pengyan Wuhan Inst Technol Sch Comp Sci & Engn Wuhan 430205 Peoples R China Wuhan Inst Technol Hubei Key Lab Intelligent Robot Wuhan 430205 Peoples R China

scene text recognition methods are broadly categorized into serial and parallel. Serial methods achieve superior accuracy but are slower in speed. Parallel methods offer faster speed but may sacrifice accuracy. Current methods struggle to strike a balance between accuracy and inference speed, particularly facing challenges in both accuracy and speed. Therefore, we propose a new scene text recognizer called SNFR. It includes a simple yet efficient decoder, Salient Neighbor Decoder (SND), which achieves high accuracy recognition with lower computational cost for attention map calculation. SND generates a neighbor matrix by selecting salient positions, which guides the generation of all the character attention maps. We also propose a text Feature Refining Module (TFRM) to capture the contextual relationship of text sequences, enhancing the overall feature representation of scene text. The experimental results demonstrate that our method achieves competitive performance on standard datasets and also shows superior performance on long text recognition.

关键词： scene text recognition Serial decoding Transformer Feature refining

来源：评论

学校读者我要写书评

暂无评论

Instruction-Guided scene text recognition

引用

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 2025年第4期47卷 2723-2738页

作者： Du, Yongkun Chen, Zhineng Su, Yuchen Jia, Caiyan Jiang, Yu-Gang Fudan Univ Sch Comp Sci Shanghai 200433 Peoples R China Beijing Jiaotong Univ Sch Comp Sci & Technol Beijing 100044 Peoples R China

Multi-modal models have shown appealing performance in visual recognition tasks, as free-form text-guided training evokes the ability to understand fine-grained visual content. However, current models cannot be trivially applied to scene text recognition (STR) due to the compositional difference between natural and text images. We propose a novel instruction-guided scene text recognition (IGTR) paradigm that formulates STR as an instruction learning problem and understands text images by predicting character attributes, e.g., character frequency, position, etc. IGTR first devises < condition,question,answer > instruction triplets, providing rich and diverse descriptions of character attributes. To effectively learn these attributes through question-answering, IGTR develops a lightweight instruction encoder, a cross-modal feature fusion module and a multi-task answer head, which guides nuanced text image understanding. Furthermore, IGTR realizes different recognition pipelines simply by using different instructions, enabling a character-understanding-based text reasoning paradigm that differs from current methods considerably. Experiments on English and Chinese benchmarks show that IGTR outperforms existing models by significant margins, while maintaining a small model size and fast inference speed. Moreover, by adjusting the sampling of instructions, IGTR offers an elegant way to tackle the recognition of rarely appearing and morphologically similar characters, which were previous challenges.

关键词： text recognition Character recognition Visualization Pipelines Computational modeling Optical character recognition Training Large language models Context modeling Benchmark testing scene text recognition instruction learning multi-modal learning character attribute

来源：评论

学校读者我要写书评

暂无评论

An adaptive multi-head self-attention coupled with attention filtered LSTM for advanced scene text recognition

引用

INTERNATIONAL JOURNAL ON DOCUMENT ANALYSIS AND recognition 2025年 1-20页

作者： Selvam, Prabu Kumar, S. N. Kannadhasan, S. SRM Inst Sci & Technol Sch Comp Tiruchirapalli Campus Tiruchirapalli 621105 India Amal Jyothi Coll Engn Dept Elect & Elect Engn Kanjirappally 686518 Kerala India Study World Coll Engn Dept Elect & Commun Engn Coimbatore India

Recognizing text embedded within a scene plays a critical role in computer vision by aiming to extract textual information from images. This task proves to be more challenging than optical character recognition due to the diverse nature of scenes and the varied shapes of the text. scene text recognition (STR) algorithms have demonstrated substantial progress using an encoder-decoder framework. The integration of diverse attention mechanisms has significantly boosted the performance of recognition models. Enhancing individual segments of the framework or employing more efficient attention mechanisms could dramatically improve overall performance. In line with this approach, we propose an advanced STR model based on the encoder-decoder architecture designed to elevate the performance of STR tasks. We introduced an adaptive multi-head self-attention mechanism that guides the model in learning when and where to utilize image features, resulting in more precise character generation. Secondly, a novel Attention-Filtered LSTM module (AFLSTM) in the decoder block presents dynamic attention into the language model. This module integrates visual attention into the LSTM output gate and tracks the sequential cell status as it changes. As a result, AFLSTM's attention takes on dynamic characteristics, enabling the currently focused visual region to direct future states. Finally, an Attention on Attention mechanism in the encoder and decoder introduces an additional attention mechanism to focus better on most contributive text regions. These three mechanisms significantly enhance the model's performance and accelerate training speed. The performance of the proposed method is benchmarked against seven established approaches, showcasing its superior performance over the majority of existing methods.

关键词： Attention mechanism Convolutional neural network Deep learning Encoder-decoder scene text recognition

来源：评论

学校读者我要写书评

暂无评论

Apvit: ViT with adaptive patches for scene text recognition

引用

DISCOVER APPLIED SCIENCES 2025年第4期7卷

作者： Zhang, Ning Li, Ce Wang, Zongshun Ma, Jialin Feng, Zhiqiang Lanzhou Univ Technol Coll Elect & Informat Engn Lanzhou 730050 Gansu Peoples R China

scene texts in nature exhibit varied colors, which serve as a significant distinguishing feature that effectively suppresses background interference. In this study, color clustering is utilized as a prior guide to group patches, enhancing their spatial relationships. Additionally, patch sizes are adaptively adjusted during training to balance speed and accuracy, while unimportant tokens and blocks in the model are pruned. We propose APViT, which modifies the ViTs model for scene text recognition requirements. It consists of three components: Sparse Patches Selection (SPS), ViT-STR, and Token Code (TC). First, SPS segments images into appropriate patches and clusters similar ones to explore diverse local patches adaptively. Second, we enhance the ViTs model specifically for scene text recognition as ViT-STR. Finally, TC prunes non-essential parts of the network based on self-attention mechanisms to accelerate performance. Consequently, our proposed APViT model outperforms state-of-the-art methods across several datasets, demonstrating its effectiveness.

关键词： Adaptive patches ViTs scene text recognition Prune

来源：评论

学校读者我要写书评

暂无评论

Exploring Out-of-Distribution scene text recognition for Driving scenes with Hybrid Test-Time Adaptation 7th

Exploring Out-of-Distribution Scene Text Recognition for Dri...

引用

7th Chinese Conference on Pattern recognition and Computer Vision

作者： Xian, Xiaoyu Qin, Jinghui Shi, Yukai Tian, Daxin Lin, Liang Beihang Univ Beijing 100080 Peoples R China CRRC Acad Co Ltd Beijing 100070 Peoples R China Guangdong Univ Technol Guangzhou 510006 Guangdong Peoples R China Sun Yat Sen Univ Guangzhou 510000 Guangdong Peoples R China

ISBN: (纸本)9789819784868;9789819784875

scene text recognition (STR) in dynamic driving scenes is important for recognizing real-world kilometer marker to facilitate the scheduling and operation of industrial scenes. For example, the location information of the train affects the safe and reliable operation of the transportation, which can be effectively determined by identifying the kilometer markers with STR technology. However, most of the existing STR models make the independent and identically distributed (i.i.d) assumption that all the training data and test data are drawn from the same data distribution. Although satisfactory performance is achieved under i.i.d assumption, existing STR models remain notoriously weak at generalization on out-of-distribution (o.o.d) data, making a system unreliable and unsafe. To validate this phenomenon, we attempt to propose a new hybrid test-time adaptation (HTTA) to improve the performance of an STR model on o.o.d test data. Previously, test-time adaptation methods are targeted at classification models and do not consider the multistep reasoning characteristic of sequence learning tasks. In HTTA, we deploy multiple semantically-reserved image augmentation and design a semantically-consistent auxiliary task to present a continual adaptation. Additionally, we construct a new Real-world Subway Kilometer Marker (RSKM) dataset for an out-of-distribution STR practice under dynamic driving scenes. We conduct extensive experiments on RSKM by embedding our HTTA into multiple classical STR methods to show the effectiveness. The experiment results show that our semantically-consistent augmentation and HTTA significantly improve the generalization performance on o.o.d STR practice.

关键词： scene text recognition Driving scenes Data Augmentation Out-of-Distribution

来源：评论

学校读者我要写书评

暂无评论

Improving scene text recognition with Counting-Aware Contrastive Learning and Attention Alignment 7th

Improving Scene Text Recognition with Counting-Aware Contras...

引用

7th Chinese Conference on Pattern recognition and Computer Vision

作者： Yang, JunJie Zhoul, Bo Zhu, Anna Wuhan Univ Technol Sch Comp Sci & Artificial Intelligence Wuhan Hubei Peoples R China Wuhan Univ Technol Chongqing Res Inst Chongqing Peoples R China

ISBN: (纸本)9789819785100;9789819785117

Contrastive learning for scene text recognition (STR) task greatly relieve the problem of relying on large scale of synthetic data or labeled data for training. Most of previous STR method using contrastive learning attempted to divide the visual features from encoder into fixed number of instances. However, fixing the number of instances may cause the mismatching between positive and negative samples in contrastive learning, especially for data with extreme geometric transformation. To tackle this problem, we introduce a very weakly labeled text length information to restrict the projection heads of the visual feature. Moreover, we found that the attention for the time-step character prediction of the same text in different shapes are different. That inspires us to think about extracting the shape-variation features of scene text, which is equal as making the attention be consistent for the same text with different shapes during decoding stage. In this paper, we propose a Counting-Aware Contrastive Learning Model (CALM) and a plug-in Attention Alignment Module (AAM) to overcome the above challenges for STR. CALM uses the text length information as guidance to determine the projection head numbers for different scene text images in the pre-training of encoder stage. AAM constrains the character attention at each time step regardless of the shape and length of text in the decoder stage. To validate the method, we synthesize the scene text data by sever transformation. Extensive experiments demonstrate the effectiveness of CALM and AAM for STR task. The comparison results also show that our proposed method achieves state-of-the-art performance on public benchmarks.

关键词： scene text recognition Attention Alignment Contrastive learning

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：