Normalized background directional feature for off-line handwritten Chinese character recognition
Fenhua Wang · Computer Engineering and Applications Journal · 2007
This paper presents an improved background directional feature extraction method.Compares to the traditional one,the new method extends the scanning directions from 4 to 8 which can detail the concavity configurations among the foreground pixels and the background pixels.Moreover,in order to reduce the influence of the stroke thickness,a normalization process is also given.Experiments on off-line handwritten Chinese amount in words show that the proposed method is more superior to the older one.