Web Multimedia Data Search Optimization Using Fibonacci-Tree Method

------------------------------------------♦----------------------------------------- I. INTRODUCTION

Multimedia data representation and retrieval is a complex process due to complex data structure and huge information matrix [1]. Many researches have been carried out in the area of multimedia data representation and information retrieval on the web. Despite differentiated by geometric data representation, multimedia objects consists of complex data structure formation during the content design process. Video as a part of multimedia component consists of multiple frames and large size of information. On the other hand, compared to video, presence of image information in any multimedia content system will always be small in size. Audio data is large in size when compared to images. Volume of text information will be more but text information occupies less data size. Hence searching multimedia component on the web is a time consuming processes [2]. Multimedia object search optimization consists of various representation models, since location of multimedia objects on the web may change its positions, orientation over a period of time [3]. Searching of image objects, animation parameters, 2D transformations, 3D positions related to multimedia data poses complex situations [4]. Searching multimedia objects like audio, video and image with compressed contents, key frame animated objects, three-dimensional motion parameters and creates abrupt challenges. Tracking and locating of multimedia object in a better way can be done by properly establishing multimedia data association of objects between various components with in the contents. For the very reason, object sequences are used as an input for the process must be differentiated. Number of methods namely sequential, binary, multiple hypotheses, nearest neighbor, graphing matching has been recommended to solve the searching problems, but all these finding suggests sequential multimedia object inputs [5]. Proper identification and righteous method of multimedia object inputs helps to establish faster and accurate search methods from a very large multimedia information system.. Hence different algorithm based multimedia data input derives different object representation and different search techniques. Further, it is observed that various tree based algorithm techniques can be superimposed on different steps involved right from image data searching, audio and video data searching, animation data locating and text data identification and retrieval.

Available online at www.ignited.in Page 2

From Fig 1, at first multimedia contents are arranged or sequenced and then forwarded for differentiation. Types of multimedia content chosen for sequencing will have the impact on the output of the subsequent step. During the multimedia content sequencing phase, multimedia objects are arranged or ordered sequentially. Since the overall availability of multimedia components on the web or on the web content management system are very large, hence data representations are also complex due to varieties of data are involved. Each content has its own multimedia objects but object position, orientation and appearance may be different. The multimedia object indexing may not be fixed, in case of multimedia objects are embedded [4]. Further multimedia object formation on the web also depends on the type of tools used for designing multimedia objects, types of tools and techniques used for representing these multimedia objects on the web. During multimedia object differentiation phase, object differentiations are being determined by differentiation technique applied on to the available types of each multimedia objects with reference to other multimedia objects on the web. The objects are then identified and sequenced based on the types, whether the objects are text, image (3D or 2D), audio or video etc.[5]. With reference to the images, the presence of images on the web may be of two-dimensional or three-dimensional [6]. Multimedia object differentiation allows user to sift multimedia content by object. In other word, multimedia object differentiation helps to search, identify and indexing of different types of multimedia objects on the web. In case of multimedia object searching phase, representation of one or more objects within the block of contents is being estimated. Further single object searching or multi object searching can be performed using various algorithms and techniques [7]. Based on interactivity, multimedia object searching algorithms are further divided into manual, automated and interactive. The beginning of entire multimedia object searching process starts with multimedia object differentiation [8]. The research identifies three basic method of multimedia object differentiation. They are sequential, random and algorithmic as shown in Figure 2.

From Fig. 2, in sequential method, multimedia objects are sequentially arranged. In most cases multimedia objects in any multimedia contents on the web will be in the mixed format. Hence categorizations or indexing of these multimedia objects and sequencing will be carried out during these phases. Without altering the sequences, the original indexed sequences are forwarded for multimedia object differentiation phase. Multimedia object comparison or data association between multimedia objects can be done on the first object set and the next succeeding object. Hence multimedia objects are traced and identified sequentially from one group of contents to another. Sequential multimedia object representation method is a slow process and to search position of object in nth object, n number of minimum operation must be performed sequentially.

Available online at www.ignited.in Page 3

Using random multimedia object sequencing method, objects are randomly arranged and then selected to forward as an input to the multimedia object differentiation process. All the multimedia object sequencing outputs are directed to object differentiation process and then to object searching. Using algorithmic method, various tree based algorithms are defined for better and efficient way of multimedia differentiation on the web. Each algorithmic method consists unique way of object sequencing methods. Hence rather than to sequencing multimedia objects in defined manner, these methods facilitate the multimedia objects arrangement as per defined algorithm and then forward it to the input of multimedia object differentiation phases The research identifies an algorithmic based technique for multimedia object searching to achieve faster results. This technique can be super imposed during multimedia object differentiation phase and multimedia object searching stages to achieve faster and accurate results.

Fibonacci-tree methods are named due to its natural characteristics and algorithmic approaches [9]. In this method, multimedia objects are differentiated using any suitable differentiation methods or Fibonacci-tree method during differentiation phase. Once the multimedia object differentiations are completed, then the multimedia object searching needs to be done. For multimedia object searching, it is possible to use any search engine algorithms or methods. But, while searching multimedia objects on the web, if the user follows the Fibonacci-tree algorithm technique, the result will be faster and accurate. Note that fibonacci-tree techniques can be superimposed during object differentiation phase or object searching phase or both. Fibonacci-tree methods are usually recommended during multimedia object differentiation phases and multimedia object searching phases for optimization.

Since the multimedia objects consists of huge data structures, searching multimedia components on the web using fibonacci-tree algorithm is more powerful, but time consuming. From the Fig. 3, M = M1,M2,M3,M4,….Mn Where M is a Multimedia Object and if it is defined as a function, then F(M12)= F(m1).F(m2) F(M12,3)= F(m12).F(m3)

…………………. …………………. ………………….

F(Mn)=F(mn-1).F(mn) Therefore F(M) ≈ F(mn-1).F(mn)

Available online at www.ignited.in Page 4

Hence using this model, it is possible to search the multimedia object by comparing one multimedia object in the content over the next multimedia object in the same contents. Indexed searching of multimedia objects can be done using any available valid method, but while searching the multimedia object, if the fibonacci-tree techniques are used, faster and accurate results can be achieved. From the above figure, Let M be the multimedia object available in the web content, then Mn = (Mn-1,Mn) where n>1 At the beginning of 1st multimedia object comparison Fn(M)= If the repeating occurrences is symmetry, then Fn(-m)=(-1)nFn(m); n M Fn(m)=(-1)n-1Fn(m); n M Hence fibonacci-tree method can be superimposed on any of multimedia object used for multimedia object differentiation and object searching procedures on the web. Various methods of multimedia object differentiation and searching based on image resolution, image attributes, video attribute, audio attribute, text representation and object searching model can be implemented using fibonacci-tree techniques, which in terms helps to achieve the better and faster results.

Multimedia data search from a huge content system require more time than any general text data search phenomenon [9[. The proposed method has been tested on various multimedia object search optimization for the web. Observe that Fibonacci-tree method search capabilities found better than general method like sequential search [10]. It has been observed that for the embedded multimedia object, optimized performance was better than external loading.

0

2.5

5

7.5

10 GeneralTree Method

Embedded

External

Tagged

When search criteria applied for various multimedia data types, it is observed that tree method search performance is superior to general sequential or random search methods.

0

2.5

5

7.5

10 GeneralTree Method

2D

3D

Text Audio Video

This paper presents a unique technique used for web multimedia data object searching for achieving faster results. The proposed algorithmic based fibonacci-tree technique can be integrated in various steps pertaining to multimedia object differentiation and retrieval procedures as well. The proposed fibonacci-tree technique can be further extended and used in indexing, searching and tracking multimedia information related to large multimedia documents, interactive multimedia contents consisting of multiple objects and multimedia authoring framework.

Available online at www.ignited.in Page 5

[1] K. Chatterjee, S.Chen “A Multimedia Data Management Approach with GeM-Tree”, Journal Of Multimedia, Vol. 5,No. 4, Academy Publisher, 2010 [2] L. Dong, H. Xian-Sheng,, W. Meng, Z.HongJiang “Boost Search Relevance for Tag-Based Social Image Retrieval”, IEEE International Conference on Multimedia and Expo (ICME), 2009 [3] W. Al-Khatib, A. Gafoor, “Semantic Modeling and Knowledge Representation in Multimedia Databases”, IEEE Transaction on Knowledge and Data Engineering (CVPR), 1999. [4] L Zhang, L Li, “3D Human Animation from 2D Monocular Data Based on Motion Trend Prediction”, 14-th International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision(WSCG),2006. [5] V.P. Sing, P Pal, “Survey of Different Types of CAPTCHA”, International Journal of Computer Science and Information Technologies, Vol-5, 2014. [6] M. Andriluka, S. Roth, B. Schiele. “Monocular 3D Pose Estimation and Tracking by Detection.”, IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2010. [7] L Feng, C Liang-Tien, P Pal, “Multimedia Web Services for an Object Tracking and Highlighting Applications”, Advances in Multimedia Modeling Lecture Notes in Computer Science Volume 4352, pp 238-247, Springer,2006 [8] de Silva, G.C.; Aizawa, K.; Arase, Y.; Xing Xie; , ”Interactive social, spatial and temporal querying for multimedia retrieval,” Content-Based Multimedia Indexing (CBMI), 2011 9th International Workshop on , vol., no., pp.7-12, 13-15 June 2011 [9] Y Wei, S Tanaka, ” Improvement of Thorup Shortest Path Algorithm by Reducing the Depth of A Component Tree” Journal of Advances in Computer Networks, Vol. 2, No. 2, June 2014 [10] K. Chatterjee, S Chen, “GeM-Tree : Towards a Generalized Multidimensional Index Structure Supporting Image and Video Retrieval in Conf. IEEE Int. Symposium on Multimedia , pp. 631-636.,2008.

Prakash Kumar U was born in Udupi, India. He pursued Master of Technology( M. Tech.) in Information Technology from Sam Higginbottom Institute of Agriculture, Technology and Sciences (Formerly Allahabad Agricultural Institute). At present he is pursuing his Ph.D. from Singhania University, Rajasthan, India. His area of research is Multimedia, Vision Computing, Computer Graphics, Multimedia Data mining, Knowledge Discovery, Digital Media design and Image Processing. He has vast experience in industry, academics and research. He involved in teaching, training, academic publications in various Universities, Premier Government Institutes, Defence Training Establishments and Public Sector Undertaking Training Institutes in India. He is currently an Associate Member of Institution of Engineers (INDIA) and also a Member of Quality Circle Forum of India.

Dr. Vijay Prakash Agrawal pursued his Masters during the year 1967 and Ph.D. from Agra University in the year 1998. He was a former Associate Professor in post-graduation department from Babu Shivnath Agrawal College; Mathura affiliated to Dr. B R Ambedkar University, Agra. During the span of 43 years of his teaching profession, Dr Vijay Prakash Agrawal contributed his services to various academic and educational endeavors.