|
Abstract.
The paper addresses the problem of improving the localization and identification of identity documents using a method based on local feature matching. Each document type is represented by a template composed of local image features, which imposes strict size constraints when scaling to tens of thousands of document types and requires effective feature selection. Existing selection algorithms do not fully consider the contribution of features to matching robustness and discriminative power. A modified feature selection algorithm is proposed to form a more informative and robust template without increasing its size. Experimental results confirm the effectiveness of the proposed method: in most test scenarios, it demonstrates higher document identification accuracy, achieving an average reduction in identification error of 10% on the public MIDV-500 dataset and 5% on MIDV-2019.
Keywords:
computer vision, document recognition, local features matching, keypoints, descriptors.
DOI 10.14357/20718632260101
EDN AQLLDZ
PP. 3-15.
References
1. Bulatov K, Arlazarov VV, Chernov T, Slavin O, Nikolaev D (2017) Smart IDReader: Document Recognition in Video Stream. In: ICDAR 2017. Institute of Electrical, Manhattan, New York, U.S., pp 39–44. https://doi.org/10.1109/ICDAR.2017.347 2. Arlazarov VV (2022) Key stages of document template processing in modern identification document recognition systems. Trudy ISA RAN (Proceedings of ISA RAS) 72:19–25. https://doi.org/10.14357/20790279220303 3. Zhu A, Zhang C, Li Z, Xiong S (2019) Coarse-to-fine document localization in natural scene image with regional attention and recursive corner refinement. International Journal on Document Analysis and Recognition (IJDAR) 22:351–360. https://doi.org/10.1007/s10032-019-00341-0 4. Awal AM, Ghanmi N, Sicre R, Furon T (2017) Complex Document Classification and Localization Application on Identity Document Images. In: 2017 14th IAPR International Conference on Document Analysis and Recognition (ICDAR). Pp 426–431. https://doi.org/10.1109/ICDAR.2017.77 5. Skoryukina NS, Shalnova EA, Arlazarov VV (2023) Method for Detecting False Responses of Localization and Identification Algorithms Using Global Features. ITiVS 28–36. https://doi.org/10.14357/20718632230403 6. Valishina NZ, Gayer AV, Skoryukina NS, Arlazarov VV (2024) Fast Keypoint Filtering for Feature-Based Identity Documents Classifcation on Complex Background. In: Osten W, Nikolaev D, Debayle J (eds) ICMV 2023. Society of Photo-Optical Instrumentation Engineers (SPIE), Bellingham, Washington 98227-0010 USA, pp 1307205-1-1307205–9. https://doi.org/10.1117/12.3023194 7. Skoryukina N, Arlazarov VV, Nikolaev DP (2020) Fast method of ID documents location and type identification for mobile and server application. In: ICDAR 2019. The Institute of Electrical, Manhattan, New York, U.S., pp 850–857. https://doi.org/10.1109/ICDAR.2019.00141 8. Regula Document Reader SDK (Accessed 23 January 2026). https://regulaforensics.com/products/documentreader-sdk/#specification 9. Skoryukina NS, Arlazarov VV, Milovzorov AN (2021) Memory Consumption Reduction for Identity Document Classification with Local and Global Features Combination. In: ICMV 2020. Society of Photo-Optical Instrumentation Engineers (SPIE), Bellingham, Washington 98227-0010 USA, pp 116051G1-116051G8. https://doi.org/10.1117/12.2587033 10. Matalov DP, Limonova EE, Skoryukina NS, Arlazarov VV (2022) Memory effcient local features descriptor for identity document detection on mobile and embedded devices. IEEE Access 11:1104–1114. https://doi.org/10.1109/ACCESS.2022.3233463 11. Bay H, Ess A, Tuytelaars T, Van Gool L (2008) SpeededUp Robust Features (SURF). Computer Vision and Image Understanding 110:346–359. https://doi.org/10.1016/j.cviu.2007.09.014 12. Trzcinski T, Christoudias M, Lepetit V (2015) Learning Image Descriptors with Boosting. IEEE Transactions on Pattern Analysis and Machine Intelligence 37:597–610. https://doi.org/10.1109/TPAMI.2014.2343961 13. Arlazarov VV, Bulatov K, Chernov T, Arlazarov VL (2019) MIDV-500: A Dataset for Identity Document Analysis and Recognition on Mobile Devices in Video Stream. Computer Optics 43:818–824. https://doi.org/10.18287/2412-6179-2019-43-5-818-824 14. Bulatov K, Matalov D, Arlazarov VV (2020) MIDV-2019: Challenges of the Modern Mobile-Based Document OCR. In: Osten W, Nikolaev D, Zhou J (eds) ICMV 2019. Society of Photo-Optical Instrumentation Engineers (SPIE), Bellingham, Washington 98227-0010 USA, pp 114332N1-114332N6. https://doi.org/10.1117/12.2558438 15. Lerouge J, Betmont G, Bres T, Stepankevich E, Bergès A (2024) DocXPand-25k: a large and diverse benchmark dataset for identity documents analysis. https://doi.org/10.48550/arXiv.2407.20662 16. Xie L, Wang Y, Guan H, Nag S, Goel R, Swamy N, Yang Y, Xiao C, Prisby J, Maciejewski R, Zou J (2024) IDNet: A Novel Identity Document Dataset via Few-Shot and Quality-Driven Synthetic Data Generation. In: 2024 IEEE International Conference on Big Data (BigData). pp 2244–2253 https://doi.org/10.1109/BigData62323.2024.10825017 17. Matalov DP, Arlazarov VV (2025) Model-driven approach to creating ID document templates for localization and classification based on a single image. ICMV 2025 - CO Journal 49:1148–1155. https://doi.org/10.18287/2412-6179-CO-1762
|
