Study On Frame And Text Recognition And Sort In Comic Pages

Comics produced a growing influence in the young consumer market in the world.Not only comics can play the role of art and literature, animation companies but alsopromote the development and sales of derivative products, such as clothing, food, crafts,etc. Animation companies have played an important role on the global economy.Meanwhile, the wild spread of PDA (Personal Digital Assistant) and e-books in recentyears make lots of e-comics services popular. E-comics is not only economic, easy toread, but also vivid and deeply favored by younger readers. For reading E-comics, thispaper do the following works：①This paper presents a method that can automatically decompose one comic pageinto multiple frames in accordance with the grammar of comics. The proposed methodwill reduce the amount of manual segmentation work. After division, comic frames canbe placed on the smaller screen of cell phone screen or PDA, one by one in order toallow simple and vivid reading even through screen with very limited size. So themethod may promote the development of e-comics. This method will identify eachframe and then arranged them in chronological order according to the story.Combined the comic grammar, this paper select a straight line between the framesas the best split line. We choose the best segmentation position by setting the divisionratio, detection line intersection numbers and other conditions, to solve the problem ofIncomplete dividing line and Mislead lines in the prospects. In the segmentation process,binary tree records the segmentation sequence. At the same time, comic frames will besorted. Experimental results show the effectiveness of the proposed method.②In order to more clearly and better reading experience, this paper presents atext location method in comic images. Sorted the comic text in the dialog can be usedfor amplification, translation and so on. The comic reading will be convenient, simple,lively.According to the Characteristics of comic text connected region, this paperpresents a method for text location in comic images. First step: After preprocessing,filtering connected regions which are too big or too small connected. Step two： thefeatures of the rest connected region be divided into two categories: region text and nontext region. Third step： text regions adjacent characteristics help locating text region.We sort text according to the comic grammar and the position of the text region. Experimental results show the effectiveness of the comic text location and sortingmethod.The innovation of this paper lies in the following points.①This paper proposed a complete set of comic frames recognition and sortingscheme. This paper proposed the best split line algorithm and comic frames methodbased on binary tree.②This paper proposed a complete set of scheme for text location and sorting incomic images. The comic text location algorithm is based on connected region and textadjacent character. And dialogs sorting is based on position of dialog boxes.