Massive network text and non-text image classification method

Abstract

The invention discloses a massive network text and non-text image classification method. Firstly a multi-scale space division network is constructed; then multi-scale image block tag information of images in a training image set is acquired and the network parameters of the multi-scale space division network are trained by using a marked training data set according to the constructed multi-scale space division network; then large-scale network images to be tested are classified by using the constructed multi-scale space division network and the trained network parameters so that the classification result of the images is acquired; and whether the images are text images is judged and the general position of the text area in the images is acquired. According to the method, the accuracy of text and non-text image classification is high and the method has high classification efficiency.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and more specifically relates to a method for classifying massive network text and non-text images. Background technique [0002] With the rapid development of television and the Internet, human society has gradually entered the information age. In the future, human economic life will be dominated by the possession, allocation, production and use of information. With the advent of the information age, more and more image and video data are disseminated in various ways, and these data contain a large amount of useful information. How to extract these useful information from these massive data will be a In the information age, it is the key to whether human beings can obtain more benefits quickly and efficiently. The current Internet provides a large amount of video and image data, and the text in these massive network video frames and network images is an extremely important source of inf...

AI Technical Summary

Problems solved by technology

However, in the mass dissemination of data, only a very small part of the images contain text, and the existing text detection and text recognition methods are limited by the speed of extracting text information in images, and it is difficult to directly extract useful information from these data. Text information, so research on text and non-text image classification algorithms has high practical significance and use value

Images