Open AccessOffline character recognition in transport logistic tables
Author(s) -
E. Yu. Bursian,
А. М. Демин
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/2131/2/022098
Subject(s) - raster graphics , skeleton (computer programming) , computer science , representation (politics) , artificial intelligence , character (mathematics) , signature (topology) , transformation (genetics) , symbol (formal) , software , pattern recognition (psychology) , computer vision , algorithm , mathematics , geometry , biochemistry , chemistry , politics , political science , law , gene , programming language
The paper proposes the improved skeleton method of handwritten characters recognition, which is based on the filtering procedure and the principle of alternating shading schemes of skeletonized area on the 4- and 8-timeslinked raster. The procedure of high-frequency filtration based on discrete real cosine transformation or discrete complex Fourier transform with automatic selection of filtration parameters makes it possible to significantly improve the image quality of handwritten symbols, in particular, to eliminate in many cases thin bridges between the areas of symbol element representation. The principle of alternating the painting schemes along the 4- and 8-timeslinked raster makes it possible to get the wave front of the skeletonized area close to a circle. In this case, the broken lines representing the branches of the skeleton graphs retain the shapes of the symbols. Numerical experiments on the construction of skeleton sets and skeleton graphs for recognizable handwritten symbols located in the cells of the tables of logistic transport problems have been performed. Software implementation of the method is proposed.