Text line detection method and device

Abstract

The invention discloses a text line detection method and device. The text line detection method comprises: a gray level decomposition step: carrying out gray level decomposition on a gray level image to acquire a text layer of the gray level image; a stroke extraction step: carrying out stroke extraction on the gray level image to acquire a stroke layer of the gray level image; a consistency verification step: verifying the consistency of connected domains of the text layer and the stroke layer, and removing the connected domain belonging to the background region of the gray level image from the text layer based on the verified result; and a text line generation step: organizing connected domains in a text layer, in which the connected domains belonging to the background region of the gray level image are removed, into a text line.. By utilizing the text line detection method and device provided by the invention, the text line which has a relative complex background and is displayed in the image or a video frame can be detected effectively and efficiently.

Description

technical field [0001] The present invention generally relates to the field of image and video processing, and more specifically relates to a text line detection method and device. Background technique [0002] In content-based image and video retrieval, text is a very powerful index. For example, in web image retrieval, many traditional image retrieval technologies use the surrounding text of the image in the webpage as the index of the image. The text embedded in the image can represent the image content more effectively and accurately than the surrounding text of the image used by traditional image retrieval techniques. Therefore, detecting, segmenting, and recognizing text in images or video frames will greatly facilitate content-based image and video retrieval. [0003] Currently existing text region detection methods can be generally divided into two categories: texture-based image text detection methods and region-based text region detection methods. Region-based t...

AI Technical Summary

Problems solved by technology

[0004] However, because the text in the image or video frame usually has a relatively complex background, at the same time, it is often difficult to be detected, segmented and recognized due to factors such as light and font, color, and position of the text, so it is effective and efficient. However, it is still a challenging task to accurately detect text regions with relatively complex backgrounds or even text lines in text regions presented in images or video frames.

Images