A Handwritten Chinese Character Recognition Method Based on Planar Shape Correction

摘要

Although the method to eliminate sample differences existed in characters by different subjects has been aroused great attention in the research of off-line handwritten Chinese recognition, almost all the known methods treat the problem by 2 1-dimensional density projections on the character bitmap image. Taken into account the fact that Chinese characters are essentially of 2-dimensional (or planar) structure, a handwritten Chinese character recognition method based on planar shape correction is introduced in this paper. Firstly the concept of planar stroke density and its design method are introduced, then the method of normalizing character image based on planar density equalization is discussed, finally some testing results are given to prove the effectiveness of this method.