An Efficient Conversion of EPIGRAPHICAL Textual Image to User Readable Text

An Efficient Conversion of EPIGRAPHICAL Textual Image to User Readable Text

1 Suresh Annamalai	2 Preethi Wilson	3 Balasathuragiri
Associate Professor & Head,	Assistant Professor	Assistant Professor

Department of Computer Science & Engineering, ASAN Memorial College of Engineering & Technology, Asan Nagar, Chengalpet 603 105.

Abstract

Epigraphy is the study of inscriptions on rocks, pillars, temple walls, copper plates and other writing material. It is one of the most fascinating and instructive studies. It deals with the art of writing, which distinguishes man from animals and provides us with an instrument for conservation and transmission of historical traditions from generation to generation. Inscriptions are the main source for reconstructing the history and culture of ancient civilizations. It serves as primary documentary evidence to establish legal, socio- cultural, literary, archaeological, and historical antiquity on the basis of engravings. The basic issues caused to the epigraphist are Paleography (letter shapes, direction, and punctuation), Non-standard language (Dialektinschriften etc), Ancient abbreviations, Chronology (ancient dating systems: eponyms, eras etc). Inscriptions are very tough to read by the common people though there is scientific support to such as Petrology and Digital enhancement. To overcome these techniques using natural language processing based on the particular knowledge generator the conversion of ancient ephigraphical text is converted and its efficiency is evaluate according to

the performance.

1. Introduction

   Fast growth of public photo and video sharing websites such as "Flickr" and "YouTube", provides a huge corpus of unstructured image and video data over the Internet. Searching and retrieving visual information from the Web, however, has been mostly limited to the use of meta-data, user-annotated tags, captions and surrounding text (e.g. the image search engine used by Google).In this paper, we present an image parsing to text description (I2T) framework that generates text descriptions in natural language based on

   understanding of image and video content. An image parsing engine parses input images into their constituent visual patterns, in a spirit similar to parsing sentences in natural language. Inscriptions, the subject of epigraphy, are of huge importance for our knowledge of the ancient world; we have thousands upon thousands of inscribed texts, ranging from small graffiti to law codes of several hundred lines. This fascinating material truly constitutes the archival sources for the ancient world (together with papyri). Our goal is to be able to locate and use inscriptions (even without knowledge of Ancient text), and also to be aware of what epigraphers actually do; only then we can critically use the editions of inscriptions which they produce. Therefore working with inscriptions, and this is very much a hands-on course where you will be locating, reading and interpreting inscribed texts yourself for the ancient era.,

2. Problem Definition

   To specify the detailed requirements of An Efficient Conversion of Epigraphical Textual Image to User Readable Text . This will explain the purpose and features of the system, what the system will do, and the constraint under which it must operate.

3. Existing System

   Over the past two decades, many researchers from the both Petrology and Digital enhancement domain have been actively investigating possible ways of retrieving the images and videos clips based on Epigraphy and achieved successfully to the certain extend perhaps, not by the common users. The major challenge is a so called semantic gap and ancient text identification, which is defined as the discrepancy between human interpretations of textual image information and those currently derived by a computer.

   1. Disadvantages

      • Time complexity

      • Decrease performance

4. Proposed System

   The work can be further extended to facilitate the inscriptions of various other scripts like Metal, Pottery, Wood, Palm leaves, Cloth, Conch shell, Mural paintings and Copper plates that were prevalent in ancient India and other non Indian regions, during the regime of various rulers.

   1. Advantages

      • Increase performance

      • Easy to read and understand

5. Architecture Design

   Describes the Overall Architecture. The Project consists of the following modules:

   • An image parsing engine

   • An And-or Graph (AoG) visual knowledge representation

   • A Semantic Web

   • A text generation engine

   1. An Image Parsing Engine

      It parses input images into parse graphs. For specific domains such as the two case study systems presented in the image frame parse is automatic. For parsing general images from the Internet for the purpose of building a large-scale image dataset, an interactive image parser is used as discussed in section.

   2. An And-or Graph (AoG) Visual Knowledge Representation

      It embodies vocabularies of visual elements including primitives, parts, objects and scenes as well as stochastic image grammar that specifies syntactic relations and semantic relations (e.g. categorical, spatial, temporal and functional relations) between these visual elements. The categorical relationships are inherited from WordNet, a lexical semantic network of English. The AoG not only guides the image parsing engine with top-down hypotheses but also serves as an ontology for mapping parse graphs into semantic representation

   3. A Semantic Web

      It interconnects different domain specific ontologies with semantic representation of parse graphs. This step helps to enrich parse graphs derived purely from visual cues with other sources of semantic information.

   4. A Text Generation Engine

      It converts semantic representations into human readable and query-able natural language descriptions.

      Fig.1. AN EFFICIENT CONVERSION OF EPIGRAPHICAL TEXTUAL IMAGE TO USER READABLE TEXT

6. Algorithm

   1. SVM Algorithm

      The Support Vector Machine (SVM) algorithm (Cortes and Vapnik, 1995) is probably the most widely used kernel learning algorithm. It achieves relatively robust pattern recognition performance using well established concepts in optimization theory.

   2. RFD Algorithm

      In this paper we present an approach based on genetic algorithms for determining optimal RDF query paths. The performance of this approach is benchmarked against the performance of a two-phase optimization algorithm.

      For more complex queries, the genetic algorithm RDFGA generally outperforms two-phase optimization in solution quality, execution time needed, and consistency in performance. Setting a time limit improves the overall performance of RDFGA compared to two-phase optimization even more.

7. Conclusion and Future Enhancements

   This paper proposes a framework that provides an end-to-end solution for parsing image and video content, extracting video event, and providing semantic and text annotation. One major contribution is the AoG visual knowledge representation. The AoG is a graphical representation for learning categorical image representations and symbolic representations simultaneously from alarge-scale image. It not only provides top-down guides during the image parsing process but also connects low-level image features with high level semantically meaningful concepts so that the parsed image can be seamlessly transformed to a semantic meta-data format and nally to a textual description.

   1. Future Enhancements

      Over the past two decades, many researchers from the both Petrology and Digital enhancement domain have been actively investigating possible ways of retrieving the images and videos clips based on Epigraphy and achieved successfully to the certain extend perhaps, not by the common users. The major challenge is a so called semantic gap and ancient text identification, which is defined as the discrepancy between human interpretations of textual image information and those currently derived by a computer. The work can be further extended to facilitate the inscriptions of various other scripts like Metal, Pottery, Wood, Palm leaves, Cloth, Conch shell, Mural paintings and Copper plates that were prevalent in ancient India and other non Indian regions, during the regime of various rulers.

8. References

[1]. Benjamin Z. Yao, Xiong Yang, Liang Lin, Mun Wai Lee and Song-Chun Zhu,I2T: Image Parsing to Text Description,,IEEE JULY 2012

[2]. Soumya A and G Hemantha Kumar, svm classifier for the prediction of era of an epigraphical script, International Journal of Peer to Peer Networks (IJP2P), April 2011

[3]. Julinda Gllavata1, Ralph Ewertp and Bernd Freisleben, A Robust Algorithm for Text Detection in Images, jan IEEE 2012

[4]. Anasuya Devi H.K, (2002), Automated Recognition of Ancient Indian Scripts, Proceedings of National workshop on Computer Vision, Graphics and Image processing, WVGIP,.

[5]. K Harish Kashyap , Bansilal , P Arun Koushik Hybrid Neural Network Architecture for Age

[6]. Manish Kumar Degraded Text recognition of Gurumukhi Script, Ph.D thesis, Thapur University, Patiala (Punjab), March 2008.

BIBLIOGRAPHY

Mr. A.Suresh., B.E.,M.Tech.,(Ph.D) works as the Associate Professor & Head of the Computer Science and Engineering Department in ASAN Memorial College of Engineering & Technology, Chengalpet,

Chennai, TamilNadu, India. He has more than 16 years of experience in teaching and his areas of specializations are Data Mining, Artificial

Intelligence, Image Processing, Neural Networks and System Software. He has published many of his research work in national and international journals & conferences and he has published one book in the name of Data structures & Algorithms in DD Publications.

Mrs. J. Preethi Wilson., B.E.,M.E.,(Ph.D) works as the Assistant Professor of the Computer Science and Engineering Department in ASAN Memorial College of Engineering & Technology, Chengalpet,

Chennai, TamilNadu, India. She has more than8 years of experience in teaching and her areas of specializations are Image Processing and Neural Networks. She has published many of her papers work in national and international conferences.

Mr. T. Balasathuragiri., B.E.,M.E.,(Ph.D) works as the Assistant Professor of the Computer Science and Engineering Department in ASAN Memorial College of Engineering & Technology, Chengalpet, Chennai, TamilNadu,

India. He has more than 3 years of experience in teaching and his areas of specializations are Image Processing and Neural Networks. He has published many of her papers work in national and international conferences.