A Multi-Oriented Text Detection Method in Natural Images Based on Linked Text Fields

Abstract

The invention discloses a method for detecting multi-directional text in natural pictures based on connecting text fields. Text fields and connections are two key steps in the detection method, which are defined as follows: text fields refer to the division of many individual text fields on a picture Multi-directional bounding box regions, which surround a text strip or part of a word; concatenation refers to connecting adjacent fields, meaning that they belong to the same word or sentence. Text fields and connections are combined to detect equidistantly at multiple scales using a fully convolutional neural network trained end-to-end. The final detection result is obtained by first connecting multiple text fields to form new regions, and then combining these new regions. Compared with the existing technology, the detection method proposed by the present invention has achieved excellent results in terms of accuracy, speed, and model simplicity. It has high efficiency and strong robustness, can overcome complex picture backgrounds, and can also detect images in images. Long text in non-Latin scripts.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and more particularly, relates to a multi-directional text detection method in natural pictures based on connected text fields. Background technique [0002] Reading text in natural images is a challenging and popular task with many practical applications in photo optical recognition, geolocation, and image retrieval. In text reading systems, text detection is to locate text regions with bounding boxes at the word level or text strip level, which is usually a very critical first step. In a sense, text detection can also be regarded as a special object detection, which takes words, characters or text strips as the detection target. [0003] Although existing techniques have achieved great success in applying object detection methods to text detection, object detection methods still have several significant deficiencies in locating text regions. First, the aspect ratio of words or text st...

AI Technical Summary

Problems solved by technology

[0003] Although existing technologies have achieved great success in applying object detection methods to text detection, there are still several obvious deficiencies in object detection methods in locating text areas.

First, the aspect ratio of words or text strips is usually much larger than that of general objects, and it is difficult for previous methods to generate bounding boxes of this ratio; second, some non-Latin texts have no gaps between adjacent words. Contains spaces, such as Chinese characters

Existing technologies can only detect words, which are not applicable when detecting this kind of text, because this kind of text without spaces cannot provide visual information to divide different words

Third, in large natural scene pictures, the text may be in any direction, but most of the existing technologies can only detect horizontal text

Images