End-to-end identification method for scene text with random shape

Abstract

The invention discloses an end-to-end identification method for a scene text with a random shape. The method comprises the steps of extracting a text characteristic through a characteristic pyramid network for generating a candidate text box by an area extracting network; adjusting the position of the candidate text box through quick area classification regression branch for obtaining more accurate position of a text bounding box; inputting the position information of the bounding box into a dividing branch, obtaining a predicated character sequence through a pixel voting algorithm; and finally processing the predicated character sequence through a weighted editing distance algorithm, finding out a most matched word of the predicated character sequence in a given dictionary, thereby obtaining a final text identification result. According to the method of the invention, the scene texts with the random shape can be simultaneously detected and identified, wherein the scene texts comprisehorizontal text, multidirectional text and curved text. Furthermore end-to-end training can be completely performed. Compared with prior art, the identification method according to the invention has advantages of obtaining advantageous effects in accuracy and versatility, and realizing high application value.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and more specifically relates to an end-to-end recognition method for arbitrary-shaped scene text. Background technique [0002] In the field of computer vision, scene text detection and recognition is a very active and challenging research direction, and many real-life applications are closely related to it, such as image-based geolocation, real-time translation, and assistance for the blind. [0003] The goal of scene text detection and recognition methods is to simultaneously detect and recognize text from natural scenes, which is divided into two tasks: detection and recognition. In most of the past studies, text detection and recognition are handled separately, that is, the first step uses the trained detector to detect the text area in the natural scene picture, and the second step uses the text area detected in the first step Input the identification module to identify and obtain ...

AI Technical Summary

Problems solved by technology

However, both of these approaches have two major drawbacks, first, neither of them can be trained fully in an end-to-end fashion

Second, these methods can only recognize horizontal or oriented text, but the shape of the text in actual scene pictures may change significantly from horizontal or oriented to curved forms

Images