Method and device for extracting text from image

Abstract

The invention provides a method and device for extracting a text from an image. The method comprises the following steps: carrying out grey level nonlinear dynamic compression on an original image to obtain an enhanced image; extracting textural characteristics and edge characteristics of the enhanced image; constructing the textural characteristics and the edge characteristics into homogeneous characteristics, and mapping the enhanced image to a homogeneous space according to the homogeneous characteristics to obtain a characteristic image; extracting a text area from the characteristic image by a text area detector; and extracting and identifying characters from the text area of the enhanced image. According to the invention, the accuracy and robustness in text detection can be effectively improved.

Description

technical field [0001] The invention belongs to the technical field of pattern recognition and computer vision, in particular to a method and device for extracting text from images. Background technique [0002] With the development of multimedia information retrieval, Internet and 3G network streaming media technology, images and videos have become the mainstream information carriers for multimedia information exchange and services. The text information in images and videos is becoming more and more important to the expression and retrieval of massive information. How to realize the automatic detection and extraction of text in images is the first step to realize image retrieval and image sensitivity discrimination based on text information. [0003] The text area in the image has features that are significantly different from the non-text area. For example, the text area has rich edges and special textures. Usually, the text area is composed of one or more lines of text, a...

AI Technical Summary

Problems solved by technology

[0007] In the methods of the prior art, the region-based method is difficult to extract accurate connected domains when the image background is complex or the image quality is poor. In addition, the formulation of heuristic rules used in the selection of text regions depends on prior knowledge , and these prior knowledge are generally difficult to obtain accurately, and the determination of many thresholds is rigid, resulting in poor robustness of the algorithm

[0008] Although the texture-based method has good versatility, the calculation is complex and the operation cost is high, and it is sensitive to the size and font of the text, resulting in low versatility of the classifier and low positioning accuracy of the text area.

This type of method also encounters difficulties when the image background contains a periodic structure texture similar to text

Images