Read the full-text of this article for free

Title: High-precision recognition method for sparsely arranged financial tables

Authors: Xiaofeng Shi; Yahui Ji; Menghui Liu

Addresses: School of Computer and Information Engineering, Fuyang Normal University, Fuyang Anhui, 236037, China; School of Computer and Information Engineering, Hefei University of Technology, Hefei Anhui, 230601, China ' School of Computer and Information Engineering, Fuyang Normal University, Fuyang Anhui, 236037, China ' School of Computer and Information Engineering, Hefei University of Technology, Hefei Anhui, 230601, China

Abstract: Table recognition and parsing is an important task in the field of information extraction. Especially in financial printing sparse layout forms, the cell spacing within the form is obvious, and the arrangement structure is relatively loose, the form frame line is complex, and the multi-line text within the cell brings a big challenge to form recognition. In order to solve this problem, an end-to-end efficient recognition algorithm for financial sparse layout forms is proposed, which realises form structure recognition and content filling by processing text semantics and structure, using dynamic difference detection algorithm, overlap-driven algorithm, and differential method. A large number of experiments have been done on 13 bank flow forms, and the results show that the mean overall accuracy is as high as 99.85%, which meets the requirements of financial-grade accuracy, and especially demonstrates strong robustness in complex borderline scenarios. The research aims to provide new theoretical support and technical reference for the field of form recognition.

Keywords: table analysis; sparse layout; end-to-end analysis; high precision.

DOI: 10.1504/IJAPR.2026.155678

International Journal of Applied Pattern Recognition, 2026 Vol.8 No.4, pp.1 - 23

Received: 15 Aug 2025
Accepted: 16 Apr 2026
Published online: 10 Aug 2026 *