(IJCSIS) International Journal of Computer Science and Information Security,Vol.
Image Retrieval with Image Tile Energy Averagingusing Assorted Color Spaces
, Sudeep D. Thepade
, Varun Lodha, Pooja Luthra, Ajoy Joseph, Chitrangada Nemani
Ph.D.Research Scholar & Associate Professor,
B.Tech (IT) Student
Information Technology Department, MPSTME,
NMIMS (Deemed-to-be University), Mumbai, India
Here the feature vector for image retrieval is composedof average energy of each tile of image for diverse number of imagetiles (like 1, 4, 9, 16, 25, 36 and 49) considered with the help of various color spaces. The paper presents exhaustive performancecomparison of 70 variants of proposed image retrieval techniqueusing ten sundry color spaces and seven image tiling methods isdone with the help of generic image database having 1000 imagesspread across 11 categories. For each proposed CBIR technique 55queries (randomly selected 5 per category) are fired on the genericimage database. To compare the performance of image retrievaltechniques, average precision and recall are computed and plottedagainst number of retrieved images. The results have shown thatRGB and HSI color spaces give the best performance for averageenergy based image retrieval across all tiles. Also it has been seen inall luminance-
chromaticity based color spaces ( Kekre’s LUV,YCbCr, YUV, YIQ and Kekre’s YCgCb) that as the number tiles
increased the overall performance also increases.Keywords: CBIR, Average Energy, Color Spaces, Image Tiling.
The large numbers of images which are being generated from avariety of sources (digital camera, digital video, scanner, theinternet etc.) have posed technical challenges for computersystems to store/transmit and index/manage image dataeffectively to make such collections easily accessible. Imagecompression deals with the challenge of storage andtransmission, where significant advancements have been made[1,4,5]. The challenge to image indexing is studied in the contextof image database [2,6,7,10,11], which has become one of thepromising and important research area for researchers from awide range of disciplines like computer vision, image processingand database areas. The thirst for better and faster imageretrieval techniques is increasing day by day. Problems withtraditional methods of image indexing have led to the rise intechniques for retrieving images on the basis of automaticallyderived features such as color, texture and shape- a technologynow referred as Content-Based Image Retrieval (CBIR). Someof the important applications for CBIR technology could beidentified as art galleries [12,14], museums, archaeology ,architecture design [8,13], geographic information systems ,weather forecast [5,22], medical imaging [5,18], trademark databases [21,23], criminal investigations [24,25], image searchon the Internet [9,19,20].A Content Based Image Retrieval (CBIR) is an interfacebetween a high level system (the human brain) and a low levelsystem (a computer). The human brain is capable of performingcomplex visual perception, but is limited in speed while acomputer is capable of limited visual capabilities at much higherspeeds. In a CBIR, features are used to represent the imagecontent. The features are extracted automatically and there is nomanual intervention, thus eliminating the dependency on humansin the feature extraction stage. These automated approaches toobject recognition are computationally expensive, difficult andtend to be domain specific. The typical CBIR system performstwo major tasks [16,17]. The first one is feature extraction (FE),where a set of features, called feature vector, is generated toaccurately represent the content of each image in the database.The second task is similarity measurement (SM), where adistance between the query image and each image in thedatabase using their feature vectors is used to retrieve the top
“closest” images [16,17,26]. For feature extraction in CBIR there
are mainly two approaches  feature extraction in spatialdomain and feature extraction in transform domain. The featureextraction in spatial domain includes the CBIR techniques basedon histograms , BTC [1,2,16], VQ [21,25,26]. The transformdomain methods are widely used in image compression, as theygive high energy compaction in transformed image [17,24]. So itis obvious to use images in transformed domain for featureextraction in CBIR . But taking transform of image is timeconsuming and also needs all images of database to be of samesize to get similar feature vectors. This limitation is overcomehere in proposed CBIR methods using average energy conceptwith help of image tiling.II.
Including RGB color space, in all ten assorted color spaces areconsidered here.
Kekre’s LUV Color Space
Kekre‟s LUV color Space is special case of Kekre Transform.
Where L gives luminance and U and V gives chromaticity valuesof color image. Positive value of U indicates prominence of redcomponent in color image and negative value of V indicatesprominence of green component. This needs the conversion of
280 http://sites.google.com/site/ijcsis/ISSN 1947-5500