Research on the Classification Calculation Based on the Local Features of Hard-Pen Chinese Characters

Abstract

At present, in the actual teaching of primary and secondary schools, the lack of full-time teachers and guidance is the biggest dilemma facing the hard-pen Chinese character education. With the rapid development of big data and artificial intelligence, computer vision technology can be extended to the field of education, which can effectively solve the problem of Chinese character classification in primary and secondary schools. In this paper, based on the expert evaluation index and the structural characteristics of Chinese characters, two models are used to complete the classification and evaluation of Chinese characters. One method uses improved HOG feature combined with PCA dimension reduction, and then uses SVM classifier to complete the classification calculation of left and right structure Chinese characters. In the other method, SSIM and Hash are used to assign different weights to the edge block, detail block and smooth block of the image respectively. Finally, the similarity of the whole image is obtained by weighting, and the evaluation and calculation of enveloping structure and single character structure of Chinese characters are completed. The experimental results show that both of the two models are effective in calculating the corresponding Chinese character structures. On this basis, a relatively perfect evaluation process of hard-pen Chinese characters in primary and secondary schools is put forward, which can solve the evaluation problem of hard-pen Chinese characters in primary and secondary schools.