Machine Learning for Texture Segmentation and Classification of Comic Image in SVG Compression

Machine Learning for Texture Segmentation and Classification of Comic Image in SVG Compression

Ray-I Chang (Department of Engineering Science and Ocean Engineering, National Taiwan University, Taipei, Taiwan), Chung-Yuan Su (Department of Engineering Science and Ocean Engineering, National Taiwan University, Taipei, Taiwan) and Tsung-Han Lin (Department of Engineering Science and Ocean Engineering, National Taiwan University, Taipei, Taiwan)
Copyright: © 2017 |Pages: 16
DOI: 10.4018/IJAMC.2017070103

Abstract

Raster comic would result in bad quality while zooming in/out. Different approaches were proposed to convert comic into vector format to resolve this problem. The authors have proposed methods to vectorize comic contents to provide not only small SVG file size and rendering time, but also better perceptual quality. However, they do not process texture in the comic images. In this paper, the authors improve their previously developed system to recognize texture elements in the comic and use these texture elements to provide better compression and faster rendering time. They propose texture segmentation techniques to partition comic into texture segments and non-texture segments. Then, the element of SVG is applied to represent texture segments. Their method uses CSG (Composite Sub-band Gradient) vector as texture descriptor and uses SVM (Support Vector Machine) to classify texture area in the comic. Then, the ACM (Active Contour Model) combining with CSG vectors is introduced to improve the segmentation accuracy on contour regions. Experiments are conducted using 150 comic images to test the proposed method. Results show that the space savings of our method is over 66% and it can utilize the reusability of SVG syntax to support comic with multiple textures. The average rendering time of the proposed method is over three times faster than the previous methods. It lets vectorized comics have higher performance to be illustrated on modern e-book devices.
Article Preview

1. Introduction

E-book market grows rapidly in recent years. With the mature development of display technology and popularity of handheld devices, readers’ habit has a significant and steady transition from printed to digital. Converting these comics from printed to digital becomes an important issue. As most of the digital comic data are still in raster formats (such as JPG, BMP, and PNG), they need to face the multiple-resolution problem which results in bad quality when zooming in/out for fitting different display devices. Conventionally, the pixel interpolation method is applied to rescale the image. It causes the degradation of image quality (such as jagged or fuzzy edges). Vector formats, such as SWF and SVG (Scalable Vector Graphics) (http://vectormagic.com/home), the sizes of files they produced are still large. They have tried to limit the file size, but image quality degraded accordingly. In Chang, Yen, and Hsu (2008), Chang and Su (2015), and Su, Chang, and Liu (2011), we applied SVG (the current version is 1.1) in raster-to-vector conversion as it is a graphic standard of W3C. For EPUB 3, SVG can be either inside an XHTML or as a standalone entity with an SVG file extension. Our methods processed comic contents to provide not only small SVG file size and rendering time but also better perceptual quality. Chang et al. adopt Autotrace to vectorize comic images, and then use the vector contour searching algorithms for removing extra spaces, combining the slope of clips, and merging similar color regions to compress the vector images. This method achieves high image quality and compression ratio. In Su et al., we recognize text elements in the comic and use these text elements to provide better compression and novel applications, such as translate comic automatically, text/content-based image search, and storyteller. Then Chang and Su improve our previously system (Chang et al.) to propose the color gradient (CG) vectorization method to identify CG regions for representing the color and the direction of CG in each region. Then, we merge neighboring regions those have the same CG vector as a large CG region and represent it by a single path of SVG with linear gradient syntax. Table 1 shows the difference main features of our previously developed system.

Table 1.
The main features of our previously developed system
ReferenceMain Features
(Chang et al., 2007)• The proposed color clustering and vector contour searching can merge similar color regions and enhance compression ratio and render speed.
(Su et al., 2011)• It can further reduce the sizes of SVG files.
• The OCR results from comic images can be translated into other languages to provide multilingual services.
• It can support text/content-based vector image search efficiently.
• It can support storyteller functionality.
(Chang & Su, 2015)• The proposed CGV can identify CG in the comic image and embed it in SVG syntax.
• The proposed post-processing can remove redundant points in SVG paths.
• The reusability of SVG syntax can further reduce the file size.
• It provides not only good perceptual quality but also small file size.

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 10: 4 Issues (2019): 1 Released, 3 Forthcoming
Volume 9: 4 Issues (2018)
Volume 8: 4 Issues (2017)
Volume 7: 4 Issues (2016)
Volume 6: 4 Issues (2015)
Volume 5: 4 Issues (2014)
Volume 4: 4 Issues (2013)
Volume 3: 4 Issues (2012)
Volume 2: 4 Issues (2011)
Volume 1: 4 Issues (2010)
View Complete Journal Contents Listing