Newspaper Layout Analysis And Recognition Based On Mobile Terminals

With the advent of the mobile Internet era, the traditional print media is gradually translating to digital media, the different preferences among intelligent end-users determine different information needs of different digital packets sections. Digital packets mobile terminal layout analysis and recognition which is the first step in character recognition system, image processing and title layout process become increasingly important, with the rapid development of intelligent terminal screen digital media.This paper presents a connectivity domain layout analysis method based on the principle of contour projection algorithm, considering the development trend of digital media, combined with the mobile terminal characteristics of digital packets and existing layout analysis and identification technology. Firstly, this algorithm expands bold word image to enhance the periodicity of contour projection histogram and facilitate the selection of communication threshold; secondly, a projection is taken from the horizontal and vertical directions of the image after the treatment, analyze the distribution location of each layout forum and spacing type of connected domain, basing on the peak of the projection histogram formed by profile projectors, trough position and width analysis; and then determine the layout properties on the basis of the unique properties of different layouts and spacing between connected domains, and merge homogeneity connected domain according to the statistical connected domain threshold, thereby get the separations of intelligent terminals picture article text area, the article title area, the image area automatically.After the layout segmentation and recognition, we use existing text recognition technology and digital image matching techniques to identify and retrieve packet header which contains main information and image area, and found a set of mobile terminals to meet the Chinese newspaper layout analysis system. Experiments show that layout analysis and recognition algorithms used in this paper could improve the robust of contour projection algorithm and robust connectivity domain analysis method, packet segmentation has ideal effect for mobile terminal shoot with good quality, it can quickly identify the location of the image header, the article title, images, and so on, which contain important information, in mobile terminal report, after combing with text recognition and image matching technology, this could expand the mobile end-user information needs to some extent.