{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,25]],"date-time":"2026-09-25T18:02:06Z","timestamp":1790359326109,"version":"4.1.0"},"reference-count":30,"publisher":"Public Library of Science (PLoS)","issue":"9","license":[{"start":{"date-parts":[[2026,9,25]],"date-time":"2026-09-25T00:00:00Z","timestamp":1790294400000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["No. 62377006"],"award-info":[{"award-number":["No. 62377006"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["www.plosone.org"],"crossmark-restriction":false},"short-container-title":["PLoS One"],"abstract":"<jats:p>Automatic transcription of braille documents remains challenging in document digitization settings, particularly under real imaging degradations such as uneven embossing, illumination variation, paper texture, skew, blur, and ghosting, which make braille dot presence uncertain and often cause error propagation from dot detection to cell encoding and final text transcription. To address this problem, we propose an uncertainty-aware structured decoding framework for Chinese braille document transcription that formulates the task as document analysis with an explicit intermediate representation rather than as a purely end-to-end black-box mapping. The method first performs robust grid fitting and structured dot analysis within braille candidate regions, and converts dot responses into calibrated probabilities. Topological consistency constraints are then imposed to recover valid six-dot cell patterns while producing dot-level and cell-level confidence estimates for risk localization. To resolve residual ambiguity efficiently, only low-confidence cells are expanded into small local candidate sets, which are rescored using contextual compatibility and local structural cost. Experiments on authentic scanned and photographed braille documents using document-level five-fold cross-validation achieved a Dot-F1 of 0.992\u2009\u00b1\u20090.002, a character error rate of 0.018\u2009\u00b1\u20090.003, and a BLEU-4 score of 74.1\u2009\u00b1\u20091.2. The results demonstrate that the proposed framework improves structural recovery stability, transcription accuracy, and sentence-level consistency, while maintaining strong robustness across document conditions and limiting additional inference overhead.<\/jats:p>","DOI":"10.1371\/journal.pone.0359339","type":"journal-article","created":{"date-parts":[[2026,9,25]],"date-time":"2026-09-25T17:43:27Z","timestamp":1790358207000},"page":"e0359339","update-policy":"https:\/\/doi.org\/10.1371\/journal.pone.corrections_policy","source":"Crossref","is-referenced-by-count":0,"title":["Uncertainty-Aware structured decoding for Chinese Braille document transcription"],"prefix":"10.1371","volume":"21","author":[{"given":"Yu","family":"Sun","sequence":"first","affiliation":[{"name":"School of Special Education, Changchun University","place":["Changchun, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0009-0482-3673","authenticated-orcid":true,"given":"Wenhao","family":"Chen","sequence":"additional","affiliation":[{"name":"School of Computer Science and Technology, Changchun University","place":["Changchun, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yihang","family":"Qin","sequence":"additional","affiliation":[{"name":"School of Computer Science and Technology, Changchun University","place":["Changchun, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wenhui","family":"Zhao","sequence":"additional","affiliation":[{"name":"School of Computer Science and Technology, Changchun University","place":["Changchun, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Haoran","family":"Sun","sequence":"additional","affiliation":[{"name":"School of Computer Science and Technology, Changchun University","place":["Changchun, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2026,9,25]]},"reference":[{"key":"pone.0359339.ref001","first-page":"150","author":"VV Mahale","year":"2025","journal-title":"A survey on braille character recognition and text-to-speech conversion systems. In: Proceedings of the International Conference on AI and Robotics. Singapore: Springer"},{"issue":"1","key":"pone.0359339.ref002","article-title":"Visual and tactile perception techniques for braille recognition","volume":"11","author":"B-S Park","year":"2023","journal-title":"Micro and Nano Syst Lett"},{"issue":"1","key":"pone.0359339.ref003","first-page":"1481","article-title":"Braille character segmentation algorithm based on Gaussian diffusion","volume":"79","author":"Z Meng","year":"2024","journal-title":"Comput Mater Contin"},{"key":"pone.0359339.ref004","doi-asserted-by":"crossref","unstructured":"Ovodov IG. Optical Braille Recognition Using Object Detection Neural Network. In: 2021 IEEE\/CVF International Conference on Computer Vision Workshops (ICCVW), 2021. 1741\u20138. https:\/\/doi.org\/10.1109\/iccvw54120.2021.00200","DOI":"10.1109\/ICCVW54120.2021.00200"},{"issue":"4","key":"pone.0359339.ref005","doi-asserted-by":"crossref","DOI":"10.1016\/j.heliyon.2024.e26155","article-title":"Fly-LeNet: A deep learning-based framework for converting multilingual braille images","volume":"10","author":"A Al-Salman","year":"2024","journal-title":"Heliyon"},{"issue":"1","key":"pone.0359339.ref006","doi-asserted-by":"crossref","first-page":"103545","DOI":"10.1016\/j.ipm.2023.103545","article-title":"A generalized ensemble approach based on transfer learning for Braille character recognition","volume":"61","author":"N Elaraby","year":"2024","journal-title":"Inf Process Manag"},{"issue":"1","key":"pone.0359339.ref007","doi-asserted-by":"crossref","first-page":"24172","DOI":"10.1038\/s41598-024-73895-7","article-title":"A deep learning approach for line-level Amharic Braille image recognition","volume":"14","author":"NA Asfaw","year":"2024","journal-title":"Sci Rep"},{"key":"pone.0359339.ref008","doi-asserted-by":"crossref","unstructured":"Li R, Liu H, Wang X, Xu J, Qian Y. Optical Braille Recognition Based on Semantic Segmentation Network with Auxiliary Learning Strategy. In: 2020 IEEE\/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2020. 2362\u20138. https:\/\/doi.org\/10.1109\/cvprw50498.2020.00285","DOI":"10.1109\/CVPRW50498.2020.00285"},{"key":"pone.0359339.ref009","doi-asserted-by":"crossref","unstructured":"Shao Z, Yu Z, Gu Y, Wang W. Braille-to-Chinese Translation System Based on Optical Braille Recognition. In: 2022 2nd Asia-Pacific Conference on Communications Technology and Computer Science (ACCTCS), 2022. 22\u20136. https:\/\/doi.org\/10.1109\/acctcs53867.2022.00013","DOI":"10.1109\/ACCTCS53867.2022.00013"},{"key":"pone.0359339.ref010","doi-asserted-by":"crossref","first-page":"102445","DOI":"10.1016\/j.displa.2023.102445","article-title":"Translating Braille into Chinese based on improved CBHG model","volume":"78","author":"T Huang","year":"2023","journal-title":"Displays"},{"key":"pone.0359339.ref011","doi-asserted-by":"crossref","first-page":"102506","DOI":"10.1016\/j.displa.2023.102506","article-title":"Pre-training model for low-resource Chinese\u2013Braille translation","volume":"79","author":"H Yu","year":"2023","journal-title":"Displays"},{"key":"pone.0359339.ref012","doi-asserted-by":"crossref","first-page":"102660","DOI":"10.1016\/j.displa.2024.102660","article-title":"Improving Braille\u2013Chinese translation with jointly trained and pre-trained language models","volume":"82","author":"T Huang","year":"2024","journal-title":"Displays"},{"key":"pone.0359339.ref013","doi-asserted-by":"crossref","first-page":"102819","DOI":"10.1016\/j.displa.2024.102819","article-title":"Enhancing Chinese\u2013Braille translation: A two-part approach with token prediction and segmentation labeling","volume":"84","author":"H Yu","year":"2024","journal-title":"Displays"},{"key":"pone.0359339.ref014","doi-asserted-by":"crossref","first-page":"126356","DOI":"10.1016\/j.eswa.2024.126356","article-title":"A non-autoregressive Chinese-Braille translation approach with CTC loss optimization","volume":"269","author":"H Yu","year":"2025","journal-title":"Expert Systems with Applications"},{"key":"pone.0359339.ref015","article-title":"Vision-Braille: an end-to-end tool for Chinese Braille image-to-text translation","author":"A Wu","year":"2024"},{"key":"pone.0359339.ref016","doi-asserted-by":"crossref","first-page":"441","DOI":"10.1007\/s10032-025-00541-x","article-title":"Enhancing music score analysis with Monte Carlo dropout: a probabilistic approach to staff-region detection","volume":"28","author":"SB Oliva-Bulpitt","year":"2025","journal-title":"Int J Doc Anal Recognit"},{"issue":"3","key":"pone.0359339.ref017","doi-asserted-by":"crossref","first-page":"357","DOI":"10.1007\/s10032-025-00543-9","article-title":"Tabular context-aware optical character recognition and tabular data reconstruction for historical records","volume":"28","author":"LG Singh","year":"2025","journal-title":"Int J Doc Anal Recognit"},{"issue":"10","key":"pone.0359339.ref018","first-page":"171","article-title":"Natural scene Braille image dataset and Braille segment detection method","volume":"49","author":"L Lu","year":"2023","journal-title":"Comput Eng"},{"issue":"8","key":"pone.0359339.ref019","doi-asserted-by":"crossref","first-page":"1415","DOI":"10.3724\/SP.J.1089.2023-00630","article-title":"Natural Scene Braille Recognition Based on SSD Framework","volume":"37","author":"D Wu","year":"2025","journal-title":"J Comput-Aided Design Comput Graphics"},{"issue":"18","key":"pone.0359339.ref020","doi-asserted-by":"crossref","first-page":"10288","DOI":"10.3390\/app151810288","article-title":"Real-time braille image detection algorithm based on improved YOLOv11 in natural scenes","volume":"15","author":"Y Sun","year":"2025","journal-title":"Appl Sci (Basel)"},{"key":"pone.0359339.ref021","doi-asserted-by":"crossref","first-page":"169357","DOI":"10.1109\/ACCESS.2021.3138240","article-title":"Deep Learning Strategy for Braille Character Recognition","volume":"9","author":"T Kausar","year":"2021","journal-title":"IEEE Access"},{"key":"pone.0359339.ref022","doi-asserted-by":"crossref","first-page":"1513","DOI":"10.1007\/s10462-023-10562-9","article-title":"A survey of uncertainty in deep neural networks","volume":"56","author":"J Gawlikowski","year":"2023","journal-title":"Artif Intell Rev"},{"key":"pone.0359339.ref023","doi-asserted-by":"crossref","first-page":"243","DOI":"10.1016\/j.inffus.2021.05.008","article-title":"A review of uncertainty quantification in deep learning: Techniques, applications and challenges","volume":"76","author":"M Abdar","year":"2021","journal-title":"Information Fusion"},{"key":"pone.0359339.ref024","unstructured":"Vasilev R, D\u2019Yakonov A. Calibration of neural networks. In: 2023. https:\/\/arxiv.org\/abs\/2303.10761"},{"issue":"14","key":"pone.0359339.ref025","doi-asserted-by":"crossref","first-page":"8073","DOI":"10.1007\/s00521-024-09505-4","article-title":"Adaptive temperature scaling for Robust calibration of deep neural networks","volume":"36","author":"SA Balanya","year":"2024","journal-title":"Neural Comput & Applic"},{"key":"pone.0359339.ref026","unstructured":"Angelopoulos AN, Bates S. A gentle introduction to conformal prediction and distribution-free uncertainty quantification. In: arXiv, 2021. https:\/\/arxiv.org\/abs\/2107.07511"},{"key":"pone.0359339.ref027","doi-asserted-by":"crossref","unstructured":"Einbinder B-S, Romano Y, Sesia M, Zhou Y. Training Uncertainty-Aware Classifiers with Conformalized Deep Learning. In: Advances in Neural Information Processing Systems 35, 2022. 22380\u201395. https:\/\/doi.org\/10.52202\/068431-1626","DOI":"10.52202\/068431-1626"},{"key":"pone.0359339.ref028","doi-asserted-by":"crossref","first-page":"861","DOI":"10.1007\/s42001-021-00149-1","article-title":"OCR with Tesseract, Amazon Textract, and Google Document AI: a benchmarking experiment","volume":"5","author":"T Hegghammer","year":"2022","journal-title":"J Comput Soc Sci"},{"key":"pone.0359339.ref029","doi-asserted-by":"crossref","first-page":"109695","DOI":"10.1016\/j.patcog.2023.109695","article-title":"End-to-End page-Level assessment of handwritten text recognition","volume":"142","author":"E Vidal","year":"2023","journal-title":"Pattern Recognition"},{"issue":"9","key":"pone.0359339.ref030","doi-asserted-by":"crossref","first-page":"12663","DOI":"10.1007\/s11063-022-10835-4","article-title":"A Comprehensive Survey on Various Fully Automatic Machine Translation Evaluation Metrics","volume":"55","author":"S Chauhan","year":"2022","journal-title":"Neural Process Lett"}],"container-title":["PLOS One"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0359339","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,25]],"date-time":"2026-09-25T17:43:33Z","timestamp":1790358213000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0359339"}},"subtitle":[],"editor":[{"given":"Shiva Ram","family":"Male","sequence":"first","affiliation":[{"name":"University of Hyderabad","place":["INDIA"]}],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,9,25]]},"references-count":30,"journal-issue":{"issue":"9","published-online":{"date-parts":[[2026,9,25]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pone.0359339","relation":{},"ISSN":["1932-6203"],"issn-type":[{"value":"1932-6203","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,25]]}}}