Optical character recognition method for deep learning model based on convolutional neural network

Abstract

The invention discloses an optical character recognition method for a deep learning model based on a convolutional neural network. The method includes the following steps: collecting data sets of common Chinese characters of different fonts, 10 Arabic numerals and 26 English letters and converting the data sets into picture formats; slightly twisting and rotating the pictures to enhance the robustness of the model, and generating a model training database; establishing the deep learning model of optical character recognition; inputting the training sets into the model, continuously optimizinga target function and learning a multi-classifier by using the convolutional neural network model through the method of supervised learning; and for a novel test sample, carrying out feature extraction on the novel test sample based on the model obtained in the previous step and obtaining a final classification result by using the model classifier. The invention provides the novel model and methodfor deep learning based on the convolutional neural network in optical character recognition. The method can be applied to general pattern classification tasks, especially text recognition. The optical character recognition model based on deep learning can significantly improve the identification correct rate of character identification.

Description

technical field [0001] The invention relates to technical fields such as computer vision, pattern recognition, and natural scene feature recognition, in particular to an optical character recognition method based on a deep learning model of a convolutional neural network. Background technique [0002] Due to its practicality in real life, optical character recognition has attracted extensive attention from scholars at home and abroad. At present, the application based on optical character recognition mainly focuses on character recognition of scanned documents. Optical character recognition also has broad application prospects in street view identifier recognition, bank ID card information recognition, classroom blackboard recognition, etc. Optical character recognition has the advantages of high efficiency and convenience. At present, a large number of research forces are constantly promoting the development of the field of optical character recognition. [0003] Usually ...

AI Technical Summary

Problems solved by technology

[0009] Aiming at the problem that the input image may be a non-standard image in the character recognition problem in the actual environment, the present invention proposes an optical character recognition method based on a neural network deep learning model, which solves the problem by placing character feature extraction and character recognition under a unified framework , so that the above two steps interact together to improve the final character recognition accuracy rate

Images