Aiming at improving the accuracy of the existing single classifier for handwritten Chinese medicine prescription recognition, a novel integrated multi-classifier method is proposed for handwritten Chinese medicine prescription recognition. Handwritten Chinese medicine prescription recognition is more complicated than single Chinese character recognition because it contains multiple lines of handwritten numbers, English letters and Chinese characters. According to the character distribution rule of handwritten Chinese medicine prescription, a novel recognition algorithm based on feature extraction algorithm is proposed. Then, a voting algorithm is proposed to integrate three basic single classifiers, which is the convolutional neural network (CNN) method, aforementioned feature based recognition algorithm, and the principal component analysis (PCA) combined with K-nearest neighbor (KNN) algorithm respectively. According to the experiment, the result shows that the recognition accuracy of the proposed integrated method in dealing with handwritten Chinese medicine prescription problem is higher than that of single classifier identification method, and the overall recognition accuracy rate can reach 99.1%.
[1]
Yoshua Bengio,et al.
Gradient-based learning applied to document recognition
,
1998,
Proc. IEEE.
[2]
Xiaoqing Ding,et al.
Handwritten character recognition using gradient feature and quadratic classifier with multiple discrimination schemes
,
2005,
Eighth International Conference on Document Analysis and Recognition (ICDAR'05).
[3]
Liu Lu,et al.
Similar Handwritten Chinese Character Recognition Based on CNN-SVM
,
2017,
ICGSP '17.
[4]
Lianwen Jin,et al.
Handwritten Chinese Character Recognition with Directional Decomposition Cellular Features
,
1998,
J. Circuits Syst. Comput..
[5]
Yan Yunhui.
Surface Defect Recognition for Steel Strips by Combining Multiple Classifiers
,
2012
.