Book Reader Interface: Image Processing Issues J. Chetan, V. Sreekanth, B. Rakesh Babu, Vamshi Ambati and C.V. Jawahar IIIT Hyderabad, Andhra Pradesh, India Abstract — Digital libraries have received wide attention as they provide access to digital information to people all over the world. We discuss how we can extend the accessibility of digital libraries to low bandwidth users and users with other varied resource constraints. We achieve it using a customizable and intelligent Book- reader that is aware of the resource constraints of the user and delivers content accordingly. We review and analyze some of the image processing issues involved in building such a book reading interface. We share the insights from developing an open source image server that addresses some of the issues. Index Terms — Digital Libraries, Image Processing. I. INTRODUCTION Digital libraries have received wide attention in the past decade. Many digitization projects have been initiated with a vision of digitizing the entire human knowledge available on the earth. The digitized content is then made available over the web for easy and continuous access of information to people all over the world. The Digital Library project (DLI) [2] of India is one such digitization project which intends to digitize large volumes of books and provide them over the internet. It started book digitization in the year 2002 with a major objective of preserving the rare and age- old collection of books related to Indian History as well as world literature. The strength of any Digital Library lies in its potential to operate over the World Wide Web and be continuously available to a wide range of audience. So it is important how digital libraries provide access to the digital information through web. In DLI, scanned images are the output from a digitization initiative and the images are stored in TIFF (Tagged Image File Format) file format. This is a widely accepted format and considered as a best practice by other digitization initiatives. Although TIFF format is good for purposes of archival, TIFF images are not often browser friendly. Also digitization usually produces images at very high possible resolution so as to preserve the full details present in the image and hence will be of huge size. This may be bandwidth intensive on a browser and may create huge delays for an end user trying to view them. For purposes of reading a book, the users may not require images of higher resolution. To improve performance and reduce time delays at the end-user, we may dynamically convert the resolution of TIFF images into other lower resolution images, suitable for viewing, and deliver them over the web. One might also convert the TIFF format into other browser friendly formats like PNG at varied resolutions to reduce the effective file size [1]. This is very useful in the case of users on a low bandwidth trying to access and read images from Digital Libraries. To make the process of browsing a required page in a book fast, thumbnails are useful as they give the visual clues and hence load fast due to their low size. In this paper we try to address the above mentioned issues to increase the flexibility of a Digital Library in delivering the content to the end-user. This in turn increases the accessibility of a Digital Library extending it to users with resource constraints. In this regard, we discuss the creation of an Image processing server for Digital Library of India project that helps in conversion of image formats (e.g.: TIFF to PNG[5] ), various resolutions (e.g.: 600 dpi to 300 dpi) and various sizes (e.g.: thumbnail). For processing of TIFF images libtiff and ImageMagick libraries are widely used. By observing the time taken by different libraries for reading, writing and manipulating TIFF images we concluded that different libraries are faster for different purposes. We have identified and configured a library for our purpose. We have developed a configurable book reading interface, which can take input preferences from the user and interacts with the Image processing server to deliver the content in formats and sizes that are suitable to the resource constraints of the user. We also discuss the creation of an open source plug-in to view TIFF images which works in MS- Windows and Unix/Linux platforms. Though many plug-ins exist they are not customized for Digital Library of India and create unwanted load on browser. Moreover they work only in single platform. The rest of paper is organized as follows. In the Section II, we discuss requirements of a user friendly book reading interface. In section III image formats preferable for digital libraries are described. In Section IV, we discuss the issues and mechanism for transferring the user required format .In Section V, we discuss how the TIFF images are presented to the user. Finally in Section VI, we discuss the Book-reader we developed and conclude in section VII.