Calligraphy font and text content synchronous identification method and system

Abstract

The invention discloses a calligraphy font and text content synchronous identification method and system. The method comprises the following steps: inputting a preprocessed calligraphy font image set into a trained convolutional neural network model based on transfer learning, and synchronously identifying Chinese calligraphy fonts to be identified and text contents, wherein the convolutional neural network is composed of seven layers; then based on the transfer learning technology, fixing network parameters of the first three layers, and using the model parameters for transferring and recognizing calligraphy fonts for recognizing Chinese character content, so that synchronous recognition of the Chinese calligraphy fonts and the character content is achieved, and the time for constructing the model is shortened. By using various efficient machine learning technologies including a back propagation algorithm, an Adam optimization algorithm based on gradient descent, SoftMax regression classification, a deep transfer learning network and the like, model training based on deep transfer learning is successfully completed, so that synchronous recognition of Chinese calligraphy fonts and text contents is accurately realized, and the time for constructing the model is shortened.

Description

technical field [0001] The invention belongs to the technical field of synchronous identification, and in particular relates to a method and system for synchronous identification of calligraphy fonts and text content. Background technique [0002] Chinese calligraphy is an ancient art developed with the development of Chinese civilization. Calligraphy exudes the charm of its ancient art all the time. Research on the use of computer technology to intelligently identify users' handwritten calligraphy works is of great significance to the promotion of traditional Chinese culture and the development of calligraphy education. [0003] At present, due to the large number of Chinese calligraphy characters and the distinct characteristics of different fonts, it is very difficult to identify calligraphy works when the fonts are unknown. Nowadays, the clustering method is commonly used to identify data, which has a high accuracy rate. However, in the image preprocessing stage, it is ...

AI Technical Summary

Problems solved by technology

[0003] At present, due to the large number of Chinese calligraphy characters and the distinct characteristics of different fonts, it is very difficult to identify calligraphy works when the fonts are unknown

Nowadays, the clustering method is commonly used to identify data, which has a high accuracy rate. However, in the image preprocessing stage, it is necessary to perform operations such as extracting the main skeleton feature of the font and removing continuous strokes, which will subjectively remove many characters. Features, especially the fonts that remove the continuous strokes will lose important feature information, which will affect the final accuracy of recognition

At the same time, in the process of clustering, it is also necessary to continuously adjust the parameters, which leads to the fact that the final recognition results still have artificial factors, which cannot satisfy the objectivity of classification.

[0004] Although the existing methods can realize the recognition of calligraphy characters based on the comparison library, in the face of the insufficient amount of experimental training data, the recognition limitations are very large, and it is difficult to apply to other fields

In addition, in the face of simultaneous recognition of multiple fonts, it is difficult to meet the experimental requirements

Therefore, the existing methods have great limitations in recognition when the amount of data is increasing rapidly and the quality of data cannot be guaranteed, and it is difficult to cope with the simultaneous recognition of large-scale calligraphy fonts and text content.

Images