Text area detection method and device

Abstract

The embodiment of the invention provides a text area detection method and device, and the method comprises the steps: obtaining a to-be-detected target image which comprises a target text; inputting the to-be-detected target image into a pre-trained semantic segmentation network model, obtaining class labels including text class labels, boundary class labels and background class labels of all pixel points in the target image, wherein the pixel points of the boundary class labels are located on the peripheries of all text areas, and therefore, different text areas can be segmented; and then, according to the pixel points of the text class labels, determining a target text region containing the target text, and correcting the target text region to obtain a rectangular text region which can facilitate text recognition such that the accuracy of text recognition is improved.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a text region detection method and device. Background technique [0002] In the field of computer vision, text detection and recognition is a research direction with great application value. Many applications in real life are closely related to it. For example, technologies such as license plate recognition and ID card information recognition have applied text detection and recognition methods. [0003] The text detection and recognition method includes two steps. First, the text region detection is performed on the image to obtain the text region in the image; second, the text recognition is performed on the text region. Among them, the quality of the text region obtained in the first step greatly affects the final text recognition accuracy. [0004] In the existing text area detection methods, the detection result is a text matrix box, which cannot handle long curved c...

AI Technical Summary

Problems solved by technology

Among them, the quality of the text area obtained in the first step greatly affects the final text recognition accuracy.

[0004] In the existing text area detection method, the detection result is a text matrix box, which cannot handle long curved strings

When the image to be detected contains continuous curved text, the detected text rectangle may frame other text areas, that is, it cannot be applied to the detection of curved text, thus affecting the accuracy of text recognition

Images