Text recognition method and device, equipment and storage medium

Abstract

The invention provides a text recognition method and device, equipment and a storage medium, relates to the technical field of image processing, and is used for improving the accuracy of text recognition. The method comprises the steps of determining a plurality of target text areas of a to-be-recognized image according to a trained text detection model; according to a trained first text recognition model, performing text recognition on the plurality of target text regions, and obtaining first text recognition results corresponding to the plurality of target text regions; according to a trained second text recognition model, performing text recognition on the plurality of target text regions, and obtaining second text recognition results corresponding to the plurality of target text regions; and determining a plurality of target text recognition results corresponding to the to-be-recognized image according to the first confidence contained in the plurality of first text recognition results and the plurality of second text recognition results.

Description

technical field [0001] The present application relates to the technical field of image processing, and provides a text recognition method, device, equipment and storage medium. Background technique [0002] In people's daily office life, we often encounter situations that require text recognition. For example, when quickly entering text in pictures, scanned text, or PDF text, since these texts cannot be directly copied and pasted, manual input is too troublesome. Therefore, the purpose of quickly entering text can be achieved through text recognition. At present, in the existing text recognition methods, the region-based convolutional neural network (Region-based Convolutional Neural Network, R-CNN) is often used to detect the text region in the image, and then the backpropagation (Back propagation) is used to detect the text region in the image. Propagation, BP) neural network to recognize text characters in text regions. [0003] However, when R-CNN is used for text dete...

AI Technical Summary

Problems solved by technology

[0003] However, when R-CNN is used for text detection, since the text spacing may be too large or too small, and the distribution direction of text lines may be in a random direction, the probability of missing or multiple detections is high, and it cannot be accurately detected. The edge of the text line is detected, which eventually leads to a lower accuracy of text detection

When using BP neural network for text recognition, due to the large number of categories of Chinese characters and the difference between simplified and traditional characters, and the punctuation marks of Chinese and English are similar, the recognition effect of Chinese characters and traditional characters is not good enough. , can only recognize a small number of specific text characters, and the applicable scenarios are few, and it is also difficult to recognize special symbols and punctuation marks, thus resulting in a low accuracy rate of text recognition and affecting the normal use of users

Images