Docstoc
EXCLUSIVE OFFER FOR DOCSTOC USERS
Try the all-new QuickBooks Online for FREE.  No credit card required.

Image Retrieval with Image Tile Energy Averaging using Assorted Color Spaces

Document Sample
Image Retrieval with Image Tile Energy Averaging using Assorted Color Spaces Powered By Docstoc
					                                                                    (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                            Vol. 9, No.3, 2011


       Image Retrieval with Image Tile Energy Averaging
                 using Assorted Color Spaces
                  Dr. H.B.Kekre1, Sudeep D. Thepade2, Varun Lodha, Pooja Luthra, Ajoy Joseph, Chitrangada Nemani 3
                         1
                           Senior Professor, 2Ph.D.Research Scholar & Associate Professor, 3B.Tech (IT) Student
                                              Information Technology Department, MPSTME,
                                       SVKM‟s NMIMS (Deemed-to-be University), Mumbai, India
                              1
                                hbkekre@yahoo.com, 2sudeepthepade@gmail.com,3ajoy.jose90@gmail.com,
                      chitrangada.nmims@gmail.com, varunlodha_4@hotmail.com, poojapoohluthra@gmail.com

Abstract— Here the feature vector for image retrieval is composed
of average energy of each tile of image for diverse number of image             A Content Based Image Retrieval (CBIR) is an interface
tiles (like 1, 4, 9, 16, 25, 36 and 49) considered with the help of             between a high level system (the human brain) and a low level
various color spaces. The paper presents exhaustive performance                 system (a computer). The human brain is capable of performing
comparison of 70 variants of proposed image retrieval technique
                                                                                complex visual perception, but is limited in speed while a
using ten sundry color spaces and seven image tiling methods is
done with the help of generic image database having 1000 images                 computer is capable of limited visual capabilities at much higher
spread across 11 categories. For each proposed CBIR technique 55                speeds. In a CBIR, features are used to represent the image
queries (randomly selected 5 per category) are fired on the generic             content. The features are extracted automatically and there is no
image database. To compare the performance of image retrieval                   manual intervention, thus eliminating the dependency on humans
techniques, average precision and recall are computed and plotted               in the feature extraction stage. These automated approaches to
against number of retrieved images. The results have shown that                 object recognition are computationally expensive, difficult and
RGB and HSI color spaces give the best performance for average                  tend to be domain specific. The typical CBIR system performs
energy based image retrieval across all tiles. Also it has been seen in         two major tasks [16,17]. The first one is feature extraction (FE),
all luminance-chromaticity based color spaces ( Kekre’s LUV,
                                                                                where a set of features, called feature vector, is generated to
YCbCr, YUV, YIQ and Kekre’s YCgCb) that as the number tiles
increased the overall performance also increases.                               accurately represent the content of each image in the database.
                                                                                The second task is similarity measurement (SM), where a
Keywords: CBIR, Average Energy, Color Spaces, Image Tiling.                     distance between the query image and each image in the
                                                                                database using their feature vectors is used to retrieve the top
                         I.    INTRODUCTION                                     “closest” images [16,17,26]. For feature extraction in CBIR there
The large numbers of images which are being generated from a                    are mainly two approaches [5] feature extraction in spatial
variety of sources (digital camera, digital video, scanner, the                 domain and feature extraction in transform domain. The feature
internet etc.) have posed technical challenges for computer                     extraction in spatial domain includes the CBIR techniques based
systems to store/transmit and index/manage image data                           on histograms [5], BTC [1,2,16], VQ [21,25,26]. The transform
effectively to make such collections easily accessible. Image                   domain methods are widely used in image compression, as they
compression deals with the challenge of storage and                             give high energy compaction in transformed image [17,24]. So it
transmission, where significant advancements have been made                     is obvious to use images in transformed domain for feature
[1,4,5]. The challenge to image indexing is studied in the context              extraction in CBIR [23]. But taking transform of image is time
of image database [2,6,7,10,11], which has become one of the                    consuming and also needs all images of database to be of same
promising and important research area for researchers from a                    size to get similar feature vectors. This limitation is overcome
wide range of disciplines like computer vision, image processing                here in proposed CBIR methods using average energy concept
and database areas. The thirst for better and faster image                      with help of image tiling.
retrieval techniques is increasing day by day. Problems with
traditional methods of image indexing have led to the rise in
techniques for retrieving images on the basis of automatically                                II.   CONSIDERED COLOR SPACES
derived features such as color, texture and shape- a technology                  Including RGB color space, in all ten assorted color spaces are
now referred as Content-Based Image Retrieval (CBIR). Some                                            considered here.
of the important applications for CBIR technology could be
identified as art galleries [12,14], museums, archaeology [3],                   A. Kekre’s LUV Color Space
architecture design [8,13], geographic information systems [5],                 Kekre‟s LUV color Space is special case of Kekre Transform.
weather forecast [5,22], medical imaging [5,18], trademark                      Where L gives luminance and U and V gives chromaticity values
databases [21,23], criminal investigations [24,25], image search                of color image. Positive value of U indicates prominence of red
on the Internet [9,19,20].                                                      component in color image and negative value of V indicates
                                                                                prominence of green component. This needs the conversion of



                                                                          280                            http://sites.google.com/site/ijcsis/
                                                                                                         ISSN 1947-5500
                                                                       (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                               Vol. 9, No.3, 2011


 RGB to LUV components. The RGB to LUV conversion matrix                     The YUV to RGB conversion matrix given in equation (6) gives
 given in equation 1 gives the L, U, V components of color image             the R, G, B components of color image for respective Y, U, V
 for respective R, G, B components.                                          components.
            L   1              1        1 R
                                                                                  R      0.7492  0.50901            1.1398 Y
            U  2             1        1. G                  (1)                 G  1.0836  0.22472               0.5876 .U          (6)
            V   0              1       1 B                                       B 0.97086          1.9729       0.000015 V
 The LUV to RGB conversion matrix given in equation 2 gives                 D. YIQ Color Space
 the R, G, B components of color image for respective L, U, V               The YIQ color space is derived from YUV color space and is
 components.                                                                optionally used by the NTSC composite color video standard.
                                                                            The ‟I‟ stands for in phase and „Q‟ for quadrature, which is the
          R  1          2        0 L/3                                     modulation method used to transmit the color information.
          G 1          1          1.U / 6                   (2)                   Y      0.299        0.587         0.144   R
          B  1          1         1 V /2                                            I  0.595716  0.274453  0.321263 . G                (7)
                                                                                    Q 0.211456  0.522591 0.31135 B
B. YCbCr Color Space
 In YCbCr color Space, Y gives luminance and Cb and Cr gives                The inter-conversion equations for YIQ to RGB color space are
 chromaticity values of color image. To get YCbCr components                given as per the equations (7) and (8).
 we need the conversion of RGB to YCbCr components. The                          R      1      0.9563             0.6210 Y
 RGB to YCbCr conversion matrix given in equation 3 gives the                   G  1  0.2721  0.6474 . I                               (8)
 Y, Cb, Cr components of color image for respective R, G, and B                  B      1       1.107            1.7046 Q
 components.
        Y      0.2989      0.5866     0.1145 R
                                                                             E. Kekre’s YCgCb Color Space
        Cb   0.1688  0.3312 0.5000 . G                     (3)            Inter-conversion equations for RGB to Kekre‟s YCgCb color
        Cr    0.5000  0.4184  0.0816 B                                     space can be given as below in equations 9 and 10.
 The YCbCr to RGB conversion matrix given in equation 4 gives                      Y    1                1            1  R
 the R, G, B components of color image for respective Y, Cb, and                   Cg  1                1           0 .G                   (9)
 Cr components.
         R    1  0.0010          1.4020 Y                                         Cb   1                0            1 B
        G  1  0.3441  0.7140 . Cb                         (4)                         R  1         1       1    Y /3
         B    1    1.7718         0.0010 Cr                                              G 1         1       0 . Cg / 2                    (10)
                                                                                         B  1         0        1 Cb / 2
 C. YUV Color Space
 The YUV model defines a color space in terms of one luminance               F. XYZ Color Space
 (brightness) and two chrominance (color) components. The YUV                Conversion equations for RGB to XYZ color space and XYZ to
 color model is used in the PAL, NTSC, and SECAM composite                   RGB can be given as given in equations 11 and 12 below.
 color video standards. Previous black-and-white systems used                      X     0.412453 0.357580 0.180423 R
 only luminance (Y) information and color information (U and V)                     Y  0.212671 0.71160 0.072169 . G
 was added so that a black-and-white receiver would still be able                                                                             (11)
 to display a color picture as a normal black and white pictures.                   Z   0.019334 0.119193 0.950227 B
 YUV models human perception of color in a different way than
 the standard RGB model used in computer graphics hardware.                         R    3.240479  1.537150  0.498535 X
 The human eye has fairly little color sensitivity: the accuracy of
 the brightness information of the luminance channel has far more                   G   0.969256 1.875992   0.041556 . Y                    (12)

 impact on the image discerned than that of the other two. The                      B    0.055648  0.204043 1.057311 Z
 RGB to YUV conversion matrix given in equation 5 gives the Y,
 U, V components of color image for respective R, G, B
 components.                                                                 G. rgb Color Space (Normalized RGB)
       Y     0.299     0.587    0.144 R                                      In order to eliminate the influence of illumination intensity, color
                                                                             information (R, G and B) can be normalized to get rgb color
       U   0.14713  0.22472 0.436 . G                       (5)           space where,
       V     0.615    0.51498 0.10001 B
                                                                                                                                               (13)




                                                                      281                             http://sites.google.com/site/ijcsis/
                                                                                                      ISSN 1947-5500
                                                                         (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                                 Vol. 9, No. 3, 2011


                                                                                  I. HSI Color Space
H. HSV Color Space                                                                To convert RGB to HSI [29,30], first we convert RGB to
The HSV stands for the Hue, Saturation and Value based on the                     „normalized rgb‟ using equations given in section 2.2.G. Each
artists (Tint, Shade, and Tone). The Value represents intensity                   normalized H, S and I are then obtained using following
of a color, which is decoupled from the color information in the                  equations.
represented image. The Hue and Saturation components are                                                                          
                                                                                              
                                                                                              
                                                                                                      1
                                                                                                         r  g   r  b      
                                                                                                                                   
                                                                                   h  cos 1 
intimately related to the way human eye perceives color resulting
                                                                                                                                    , h  0,   , forb  g
                                                                                                      2
in image processing algorithms with physiological basis.                                                                         1                                   (20)
                                                                                                        2  r  g g  b  2
Conversion formula from RGB to HSV is as follows.                                              r  g                          
                                                                                                                                
                                                                                                                                     
                                                          (14)                                     
                                                                                                   
                                                                                                           1
                                                                                                              r  g   r  b   
                                                                                  h  2  cos  1        2
                                                                                                                                       , h  ,2  , for, b  g    (21)
                                                                                                                                 1 
                                                                                                                                   2
                                                                                                    r  g   r  g g  b  
                                                                                                              2
                                                                                                                                 
                                                          (15)
                                                                                                 s  1  3. minr, g , b            , s  0,1                      (22)


                                                          (16)
                                                                                               i  ( R  G  B) /(3.255)                , i  0,1                   (23)

Conversion from HSV space to RGB space is more complex.                           For convenience h,s and i values are converted in the ranges of
And, given to the nature of the hue information, we will have a                   [0,360],[0,100] and [0,255] respectively using following
different formula for each sector of the color triangle.                          equation 24.
                                                                                       H  h 180 / , S  s 100, I  i  255                                       (24)
 Red-Green Sector:
                   for
                                                                 (17)                                      III.     IMAGE TILING [24]
                                                                                  Tiling of an image is basically dividing an image into different
                                                                                  equal sized, non overlapping quadrants for feature extraction.
                                                                                  Here seven assorted image tiling techniques are applied on
                                                                                  images for feature extraction per colour space in proposed CBIR
                                                                                  methods.
 Green-Blue Sector:
                  for
                                                                 (18)                             IV.      PROPOSED CBIR TECHNIQUES
                                                                                  Energy averaging could be defined as saverage of squared values
                                                                                  of pixels of the respective image tile. Here average energy
                                                                                  technique is applied along with tiling for feature vector
                                                                                  generation. Feature vector size for respective image tiling
 Blue-Red Sector:                                                                 method is shown in table 1.
                  for
                                                                 (19)




                                              1-Tile       4-Tile        9-Tile        16-Tile          25-Tile      36-Tile         49-Tile
                    Image Tiling Method
                                              (1x1)        (2x2)         (3x3)          (4x4)            (5x5)        (6x6)           (7x7)
                        Number of Tiles          1               4            9           16              25            36              49
                    Feature vector Size          3           12           27              48              75            108            147
                                      Table 1 Feature vector size for respective image tiling method considered




                                                                        282                                       http://sites.google.com/site/ijcsis/
                                                                                                                  ISSN 1947-5500
                                                                                                    (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                                                            Vol. 9, No.3, 2011
                                                         V.   IMPLEMENTATION                                      definitions for these two measures are given by following
The implementation of the CBIR techniques is done in                                                              equations.
MATLAB 7.0 using a computer with Intel Core 2 Duo                                                                                 Number _ of _ relevant_ images_ retrieved
Processor T8100 (2.1GHz) and 2 GB RAM. The CBIR                                                                     Pr ecision
                                                                                                                                   Total _ number_ of _ images_ retrieved           (25)
techniques are tested on the image database [15] of 1000
variable size images spread across 11 categories of human
being, animals, natural scenery and manmade things. The                                                                            Number _ of _ relevant_ images_ retrieved
                                                                                                                  Re call 
categories and distribution of the images are Tribes (85),                                                                    Total _ number_ of _ relevent _ images_ in _ database
                                                                                                                                                                                  (26)
Buses (99), Beaches (99), Dinosaur (99), Elephants (99),
Roses (99), Horses (99), Mountains (61), Airplanes (100),
Monuments (99) and Sunrise (61). Figure 2 gives the
sample images from generic image database.                                                                                      VI.     RESULTS AND DISCUSSION
                                                                                                                  For testing the performance of each proposed CBIR
                                                                                                                  technique randomly selected five images from each
                                                                                                                  category are fired on the database as queries. The average
                                                                                                                  precision and average recall are computed by grouping the
                                                                                                                  number of retrieved images. Figure 2 gives performance
                                                                                                                  comparison of assorted color spaces with average energy
                                                                                                                  technique based on image tiling. It is seen that as the
                                                                                                                  image tiling increases so does the performance increase.
                                                                                                                  The overall best performance is given by HSI color space
                                                                                                                  in tile-9 (3x3). As far as the color spaces go, the best
                                                                                                                  results for each of the tiling methods is given by RGB and
                                                                                                                  HIS color spaces. Also it has been noted that as the no of
                                                                                                                  tiles goes on increasing in all luminance-chromaticity
        Figure 1: Sample images of Generic Image Database [Image                                                  based color spaces like Kekre‟s LUV, YCbCr, YUV, YIQ
          database contains total 1000 images with 11 categories]
                                                                                                                  and Kekre‟s YCgCb, the overall performance also
To assess the retrieval effectiveness, we have used the                                                           increases
precision and recall as statistical comparison parameters
[1,2] for the proposed CBIR techniques. The standard



                                                                                    1X1     2X2      3X3      4X4      5X5        6X6     7X7

                                                  0.42
  Crossover Point Value of Average Precision




                                                  0.41
                                                   0.4
                                                  0.39
                                                  0.38
                                                  0.37
                                                  0.36
                                                  0.35
             and Average Recall




                                                  0.34
                                                  0.33
                                                  0.32
                                                  0.31
                                                   0.3
                                                  0.29
                                                  0.28
                                                  0.27
                                                  0.26
                                                  0.25
                                                  0.24
                                                  0.23
                                                              RGB         HSI         XYZ         HSV          rgb        YCgCb          LUV         YUV          YIQ         YCbCr
                                                                                Assorted Color Spaces Considered for Amendment of Energy Average

                                               Figure 2: Comparison of considered image tiling methods in proposed CBIR techniques for individual color space consideration




                                                                                                                283                                  http://sites.google.com/site/ijcsis/
                                                                                                                                                     ISSN 1947-5500
                                                                                                          (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                                                                  Vol. 9, No.3, 2011

                                                                        RGB     HSI         XYZ   HSV      rgb       YCgCb     LUV      YUV       YIQ      YCbCr

                                                    0.42
   Crossover Point Value of Average Precision and



                                                    0.41
                                                     0.4
                                                    0.39
                                                    0.38
                                                    0.37
                                                    0.36
                                                    0.35
                   Average Recall




                                                    0.34
                                                    0.33
                                                    0.32
                                                    0.31
                                                     0.3
                                                    0.29
                                                    0.28
                                                    0.27
                                                    0.26
                                                    0.25
                                                    0.24
                                                    0.23
                                                                  1X1                 2X2               3X3              4X4               5X5                6X6                 7X7
                                                                                  Assorted Color Spaces Considered for Amendment of Energy Average



                                                           Figure 3: Comparison of color spaces considered in proposed CBIR methods for individual tiling consideration




Figure 3 shows the performance comparison of image tiling                                                                                 VIII.    REFERENCES
with average energy technique based on assorted color spaces.                                                         [1] H.B.Kekre, Sudeep D. Thepade, “Boosting Block Truncation
For HSV color space, there is a sudden increase in the                                                                    Coding using Kekre’s LUV Color Space for Image
performance as tiling is applied on the images. For non-                                                                  Retrieval”, WASET International Journal of Electrical,
luminance based color spaces like RGB, HIS, XYZ, HSV and                                                                  Computer and System Engineering (IJECSE), Volume 2,
normalized RGB, the performance increases till tile-9 and then                                                            Number 3, pp. 172-180, Summer 2008. Available online at
it starts decreasing. In luminance-chromaticity color spaces                                                              http://www.waset.org/ijecse/v2/v2-3-23.pdf.
like Kekre‟s YCbCg, Kekre‟s LUV, YUV, YIQ, YCbCr, the
                                                                                                                      [2] H.B.Kekre, Sudeep D. Thepade, “Image Retrieval using
performance increases after tile-16. The improvement in
                                                                                                                          Augmented Block Truncation Coding Techniques”, ACM Int.
performance of image retrieval with increasing number of
                                                                                                                          Conf. on Advances in Computing, Comm. & Control
.tiles is observed upto certain extent, if the number of tiles
                                                                                                                          (ICAC3-2009), 23-24 Jan 2009, FCRCE, Mumbai. Is
increased further the performance deteriorates.
                                                                                                                          uploaded on online ACM portal.
                                                                                                                      [3] H.B.Kekre, Sudeep D. Thepade, “Scaling Invariant Fusion of
                                                             VII. CONCLUSION                                              Image Pieces in Panorama Making and Novel Image Blending
So far many CBIR techniques have been proposed, but still the                                                             Technique”, International Journal on Imaging (IJI),
researchers are craving for better and faster image retrieval                                                             www.ceser.res.in/iji.html, Volume 1, No. A08, pp. 31-46,
solutions. The paper presented the exhaustive comparison of                                                               Autumn 2008.
image retrieval techniques based on average energy technique                                                          [4] Hirata K. and Kato T. “Query by visual example – content-
using tiling on 10 assorted color spaces. The techniques are                                                              based image retrieval”, In Proc. of Third International
tested on image database of 1000 generic images spread across                                                             Conference on Extending Database Technology, EDBT’92,
11 different categories .The overall best performance is given                                                            1992, pp 56-71
by HSI color space in tile-9. As far as the colorspaces go, the                                                       [5] H.B.Kekre, Sudeep D. Thepade, “Rendering Futuristic Image
best results throughout all the tiling is given by RGB and HSI.                                                           Retrieval System”, National Conference on Enhancements in
Also it has been noted that as the no of tiles goes on increasing                                                         Computer, Communication and Information Technology,
in all luminance chromaticity based color spaces like Kekre‟s                                                             EC2IT-2009, 20-21 Mar 2009, K.J.S. COE, Vidyavihar,
LUV, YCbCr, YUV, YIQ and Kekre‟s YCgCb, the overall                                                                       Mumbai-77.
performance also increases.




                                                                                                               284                                http://sites.google.com/site/ijcsis/
                                                                                                                                                  ISSN 1947-5500
                                                                  (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                            Vol. 9, No.3, 2011
[6]    Minh N. Do, Martin Vetterli, “Wavelet-Based Texture                        International Journal of Computer Science and Network
       Retrieval Using Generalized Gaussian Density and Kullback-                 Security (IJCSNS),Volume:10, Number 1, January 2010,
       Leibler Distance”, IEEE Transactions On Image Processing,                  (ISSN: 1738-7906) Available at www.IJCSNS.org.
       Volume 11, Number 2, pp.146-158, February 2002.                       [18] H.B.Kekre, Sudeep D. Thepade, Archana Athawale, Anant
[7]    B.G.Prasad, K.K. Biswas, and S. K. Gupta, “Region –based                   Shah, Prathmesh Verlekar, Suraj Shirke, “Walsh Transform
       image retrieval using integrated color, shape, and location                over Row Mean and Column Mean using Image
       index”, International Journal on Computer Vision and Image                 Fragmentation and Energy Compaction for Image Retrieval”,
       Understanding Special Issue: Colour for Image Indexing and                 International Journal on Computer Science and Engineering
       Retrieval, Volume 94, Issues 1-3, April-June 2004, pp.193-                 (IJCSE),Volume 2S, Issue1, January 2010, Available online at
       233.                                                                       www.enggjournals.com/ijcse.
[8]    H.B.Kekre, Sudeep D. Thepade, “Creating the Color                     [19] H.B.Kekre, Sudeep D. Thepade, “Image Retrieval using
       Panoramic View using Medley of Grayscale and Color Partial                 Color-Texture Features Extracted from Walshlet Pyramid”,
       Images ”, WASET International Journal of Electrical,                       ICGST International Journal on Graphics, Vision and Image
       Computer and System Engineering (IJECSE), Volume 2, No.                    Processing (GVIP), Volume 10, Issue I, Feb.2010, pp.9-18,
       3,      Summer         2008.      Available       online     at            Available                                                online
       www.waset.org/ijecse/v2/v2-3-26.pdf.                                       www.icgst.com/gvip/Volume10/Issue1/P1150938876.html
[9]    Stian Edvardsen, “Classification of Images using color, CBIR          [20] H.B.Kekre, Sudeep D. Thepade, “Color Based Image
       Distance Measures and Genetic Programming”, Ph.D. Thesis,                  Retrieval using Amendment Block Truncation Coding with
       Master of science in Informatics, Norwegian university of                  YCbCr Color Space”, International Journal on Imaging (IJI),
       science and Technology, Department of computer and Info.                   Volume 2, Number A09, Autumn 2009, pp. 2-14. Available
       science, June 2006.                                                        online at www.ceser.res.in/iji.html (ISSN: 0974-0627).
[10]   H.B.Kekre, Tanuja Sarode, Sudeep D. Thepade, “DCT                     [21] H.B.Kekre, Tanuja Sarode, Sudeep D. Thepade, “Color-
       Applied to Row Mean and Column Vectors in Fingerprint                      Texture Feature based Image Retrieval using DCT applied on
       Identification”, In Proceedings of International Conference on             Kekre’s Median Codebook”, International Journal on
       Computer Networks and Security (ICCNS), 27-28 Sept. 2008,                  Imaging (IJI), Volume 2, Number A09, Autumn 2009,pp. 55-
       VIT, Pune.                                                                 65. Available online at www.ceser.res.in/iji.html (ISSN:
[11]   Zhibin Pan, Kotani K., Ohmi T., “Enhanced fast encoding                    0974-0627).
       method for vector quantization by finding an optimally-               [22] H.B.Kekre, Sudeep D. Thepade, Akshay Maloo “Performance
       ordered Walsh transform kernel”, ICIP 2005, IEEE                           Comparison for Face Recognition using PCA, DCT
       International Conference, Volume 1, pp I - 573-6, Sept. 2005.              &WalshTransform of Row Mean and Column Mean”, ICGST
[12]   H.B.kekre, Sudeep D. Thepade, “Improving „Color to Gray                    International Journal on Graphics, Vision and Image
       and Back ‟ using Kekre’s LUV Color Space”, IEEE                            Processing (GVIP), Volume 10, Issue II, Jun.2010, pp.9-18,
       International Advanced Computing Conference 2009                           Available                                                online
       (IACC’09), Thapar University, Patiala, INDIA, 6-7 Mar                      http://209.61.248.177/gvip/Volume10/Issue2/P1181012028.p
       2009. Available at IEEE Xplore.                                            df..
[13]   H.B.Kekre, Sudeep D. Thepade, “Image Blending in Vista                [23] H.B.Kekre, Sudeep D. Thepade, “Improving the Performance
       Creation using Kekre's LUV Color Space”, SPIT-IEEE                         of Image Retrieval using Partial Coefficients of Transformed
       Colloquium and International Conference, Sardar Patel                      Image”, International Journal of Information Retrieval, Serials
       Institute of Technology, Andheri, Mumbai, 04-05 Feb 2008.                  Publications, Volume 2, Issue 1, 2009, pp. 72-79 (ISSN:
[14]   H.B.Kekre, Sudeep D. Thepade, “Color Traits Transfer to                    0974-6285)
       Grayscale Images”, In Proc.of IEEE First International                [24] H.B.Kekre, Sudeep D. Thepade, Archana Athawale, Anant
       Conference on Emerging Trends in Engg. & Technology,                       Shah, Prathmesh Verlekar, Suraj Shirke, “Performance
       (ICETET-08), G.H.Raisoni COE, Nagpur, INDIA. Uploaded                      Evaluation of Image Retrieval using Energy Compaction and
       on online IEEE Xplore.                                                     Image Tiling over DCT Row Mean and DCT Column Mean”,
[15]   http://wang.ist.psu.edu/docs/related/Image.orig (Last referred             Springer-International Conference on Contours of Computing
       on 23 Sept 2008)                                                           Technology (Thinkquest-2010), Babasaheb Gawde Institute
[16]   H.B.Kekre, Sudeep D. Thepade, “Using YUV Color Space to                    of Technology, Mumbai, 13-14 March 2010, The paper will
       Hoist the Performance of Block Truncation Coding for Image                 be uploaded on online Springerlink.
       Retrieval”, IEEE International Advanced Computing                     [25] H.B.Kekre, Tanuja K. Sarode, Sudeep D. Thepade, Vaishali
       Conference 2009 (IACC09), Thapar University, Patiala,                      Suryavanshi,“Improved Texture Feature Based Image
       INDIA, 6-7 Mar 2009.                                                       Retrieval using Kekres Fast Codebook Generation
[17]   H.B.Kekre, Sudeep D. Thepade, Archana Athawale, Anant                      Algorithm”, Springer-International Conference on Contours
       Shah, Prathmesh Verlekar, Suraj Shirke,“Energy Compaction                  of Computing Technology (Thinkquest-2010), Babasaheb
       and Image Splitting for Image Retrieval using Kekre                        Gawde Institute of Technology, Mumbai, 13-14 March 2010.
       Transform over Row and Column Feature Vectors”,




                                                                      285                               http://sites.google.com/site/ijcsis/
                                                                                                        ISSN 1947-5500
                                                                (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                        Vol. 9, No.3, 2011
  [26] H.B.Kekre, Tanuja K. Sarode, Sudeep D. Thepade, “Image           „Manshodhan 2010‟ and Best Paper Award at Springer Int. Conf.
       Retrieval by Kekre’s Transform Applied on Each Row of            ICCCT-2010.
       Walsh Transformed VQ Codebook”, (Invited), ACM-
                                                                        Ajoy Joseph, Chitrangada Nemani, Pooja Luthra, Varun Lodha
       International Conference and Workshop on Emerging Trends         are currently pursuing B.Tech. (IT) from MPSTME, SVKM‟s
       in Technology (ICWET 2010),Thakur College of Engg. And           NMIMS University, Mumbai. There area of interest is Content Based
       Tech., Mumbai, 26-27 Feb 2010, Invited paper at ICWET            Image Retrieval in Image Processing
       2010 available at online ACM Portal.
  [27] H.B.Kekre, Sudeep D. Thepade, Adib Parkar, “A Comparison
       of Kekres Fast Search and Exhaustive Search for various Grid
       Sizes used for Colouring a Greyscale Image”, 2nd
       International Conference on Signal Acquisition and
       Processing (ICSAP 2010), IACSIT, Bangalore, pp. 53-57, 9-
       10 Feb 2010, The paper is uploaded on online IEEE Xplore.


                IX.    AUTHOR BIOGRAPHIES
                   Dr. H. B. Kekre has received B.E. (Hons.) in
                  Telecomm. Engineering. from Jabalpur University in
                  1958, M.Tech (Industrial Electronics) from IIT
                  Bombay in 1960, M.S.Engg. (Electrical Engg.) from
                  University of Ottawa in 1965 and Ph.D. (System
Identification) from IIT Bombay in 1970 He has worked as
Faculty of Electrical Engg. and then HOD Computer Science and
Engg. at IIT Bombay. For 13 years he was working as a professor and
head in the Department of Computer Engg. at Thadomal Shahani
Engineering. College, Mumbai. Now he is Senior Professor at
MPSTME, SVKM‟s NMIMS University. He has guided 17 Ph.Ds,
more than 100 M.E./M.Tech and several B.E./B.Tech projects. His
areas of interest are Digital Signal processing, Image Processing and
Computer Networking. He has more than 320 papers in National /
International Conferences and Journals to his credit. He was Senior
Member of IEEE. Presently He is Fellow of IETE and Life Member
of ISTE Recently ten students working under his guidance have
received best paper awards and two have been conferred Ph.D.
degree of SVKM‟s NMIMS University. Currently 10 research
scholars are pursuing Ph.D. program under his guidance.

                 Sudeep D. Thepade has Received B.E.(Computer)
                degree from North Maharashtra University with
                Distinction in 2003. M.E. in Computer Engineering
                from University of Mumbai in 2008 with Distinction,
                currently pursuing Ph.D. from SVKM‟s NMIMS
                University, Mumbai. He has about than 08 years of
experience in teaching and industry. He was Lecturer in Dept. of
Information Technology at Thadomal Shahani Engineering College,
Bandra(w), Mumbai for nearly 04 years. Currently working as
Associate Professor in Computer Engineering at Mukesh Patel
School of Technology Management and Engineering, SVKM‟s
NMIMS University, Vile Parle(w), Mumbai, INDIA. He is member
of International Association of Engineers (IAENG) and International
Association of Computer Science and Information Technology
(IACSIT), Singapore. He is reviewer for many international journals
and in the international advisory panel for many international
conferences. He has worked as member of International Advisory
Committee for many International Conferences. His areas of interest
are Image Processing and Computer Networks. He has about 100
papers in National/International Conferences/Journals to his credit
with a Best Paper Award at Int. Conference SSPCCIN-2008, Second
Best Paper Award at ThinkQuest-2009 National Level faculty paper
presentation competition, second award for research project at




                                                                        286                          http://sites.google.com/site/ijcsis/
                                                                                                     ISSN 1947-5500