Image Retrieval using Shape Texture Patterns generated from Walsh-Hadamard Transform and Gradient Image Bitmap by ijcsiseditor

VIEWS: 503 PAGES: 7

More Info
									                                                             (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                     Vol. 08, No.09, 2010

       Image Retrieval using Shape Texture Patterns
      generated from Walsh-Hadamard Transform and
                  Gradient Image Bitmaps
                                    Dr. H.B.Kekre1, Sudeep D. Thepade2, Varun K. Banura3
                     1
                         Senior Professor, 2Ph.D.Research Scholar & Associate Professor, 3B.Tech (CE) Student
                                             Computer Engineering Department, MPSTME,
                                     SVKM‟s NMIMS (Deemed-to-be University), Mumbai, India
                            1
                             hbkekre@yahoo.com, 2sudeepthepade@gmail.com,3varunkbanura@gmail.com

Abstract— The theme of the work presented here is gradient               [6], architecture design [11,16], geographic information
mask texture based image retrieval techniques using image                systems [8], weather forecast [8,25], medical imaging [8,21],
bitmaps and texture patterns generated using Walsh-Hadamard              trademark databases [24,26], criminal investigations [27,28],
transform. The shape of the image is extracted by using three            image search on the Internet [12,22,23]. The paper attempts to
different gradient operators (Prewitt, Robert and Sobel) with            provide better and faster image retrieval techniques.
slope magnitude method followed by generation of bitmap of the
shape feature extracted. This bitmap is then compared with the           A. Content Based Image Retrieval
different texture patterns namely ‘4-pattern’, ‘16-pattern’ and
                                                                             For the first time Kato et.al. [7] described the experiments
‘64-pattern’ generated using Walsh-Hadamard transform matrix
to produce the feature vector as the matching number of ones
                                                                         of automatic retrieval of images from a database by colour and
and minus ones per texture pattern. The proposed content based           shape feature using the terminology content based image
image retrieval (CBIR) techniques are tested on a generic image          retrieval (CBIR). The typical CBIR system performs two major
database having 1000 images spread across 11 categories. For             tasks [19,20] as feature extraction (FE), where a set of features
each proposed CBIR technique 55 queries (randomly selected 5             called feature vector is generated to accurately represent the
per image category) are fired on the image database. To compare          content of each image in the database and similarity
the performance of image retrieval techniques average precision          measurement (SM), where a distance between the query image
and recall of all the queries per image retrieval technique are          and each image in the database using their feature vectors is
computed. The results have shown improved performance                    used to retrieve the top “closest” images [19,20,29].
(higher precision and recall values of crossover points) with the
proposed methods compared to the mask-shape based image                       For feature extraction in CBIR there are mainly two
retrieval techniques. Further the performance of proposed image          approaches [8] feature extraction in spatial domain and feature
retrieval methods is enhanced using even image part. In the              extraction in transform domain. The feature extraction in
discussed image retrieval methods, the combination of original           spatial domain includes the CBIR techniques based on
and even image part for 4-pattern texture with shape masks               histograms [8], BTC [4,5,19], VQ [24,28,29]. The transform
generated using Robert gradient operator gives the highest               domain methods are widely used in image compression, as they
crossover point of precision and recall indicating better                give high energy compaction in transformed image [20,27]. So
performance.                                                             it is obvious to use images in transformed domain for feature
                                                                         extraction in CBIR [26]. But taking transform of image is time
    Keywords- CBIR, Gradient operators,         Walsh-Hadamard           consuming. Spatial feature based CBIR methods are given in
transform, Texture, Pattern, Bitmap.                                     [30] as mask-shape CBIR and mask-shape BTC CBIR. The
                                                                         proposed CBIR methods are further attempting to improve the
                       I.    INTRODUCTION
                                                                         performance of these shape based image retrieval with help of
    Today the information technology experts are facing                  shape texture patterns. Here the query execution time is further
technical challenges to store/transmit and index/manage image            reduced by decreasing the feature vector size further and
data effectively to make easy access to the image collections of         making it independent of image size. Many current CBIR
tremendous size being generated due to large numbers of                  systems use the Euclidean distance [4-6,11-17] on the extracted
images generated from a variety of sources (digital camera,              feature set as a similarity measure. The Direct Euclidian
digital video, scanner, the internet etc.). The storage and              Distance between image P and query image Q can be given as
transmission is taken care of by image compression [4,7,8].              equation 1, where Vpi and Vqi are the feature vectors of image
The image indexing is studied in the perspective of image                P and Query image Q respectively with size „n‟.
database [5,9,10,13,14] as one of the promising and important
research area for researchers from disciplines like computer                                         n

vision, image processing and database areas. The hunger of                                  ED      (Vpi  Vqi )
                                                                                                    i 1
                                                                                                                     2
                                                                                                                                           (1)
superior and quicker image retrieval techniques is increasing
day by day. The significant applications for CBIR technology
could be listed as art galleries [15,17], museums, archaeology




                                                                    76                              http://sites.google.com/site/ijcsis/
                                                                                                    ISSN 1947-5500
                                                                  (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                          Vol. 08, No.09, 2010
                  II.    EDGE DETECTION MASKS                               column at a time to get one pattern). The texture patterns
    Edge detection is a very important in image analysis. As                obtained are orthogonal in nature.
the edges give idea about the shapes of objects present in the                       Figure 1(a) shows a 2X2 Walsh-Hadamard matrix.
image so they are useful for segmentation, registration, and                The four texture patterns generated using this matrix are
identification of objects in a scene. An edge is a jump in                  shown in figure 1(b). Similarly figure 2(b) shows first four
intensity. An ideal edge is a discontinuity (i.e., a ramp with an           texture patterns (out of total 16) generated using 4X4 Walsh-
infinite slope). The first derivative assumes a local maximum               Hadamard matrix shown in figure 2(a).
at an edge. The various gradient operators [13] used for edge
extraction are Prewitt, Roberts and Sobel.
                III.    SLOPE MAGNITUDE METHOD
    The problem with edge extraction using gradient operators
is detection of edges in only either horizontal or vertical
directions. Shape feature extraction in image retrieval requires
the extracted edges to be connected in order to reflect the
boundaries of objects present in the image. Slope magnitude
method is used along with the gradient operators (Prewitt,
Robert and Sobel) to extract the shape features in form of
connected boundaries. The process of applying the slope
magnitude method is given as follows. First one needs to                            1(a). 2x2 Walsh-Hadamard transform matrix
convolve the original image with the Gx mask to get the x
gradient and Gy mask to get the y gradient of the image. Then
the individual squares of both are taken. Finally the two
squared terms are added and square root of this sum is taken as
given in equation 2.

                  ���� =        2      2
                           �������� + ��������                      (2)


        IV.   TEXTURE PATTERNS USING WALSH-HADAMARD
                     TRANSFORM MATRIX
    Walsh transform matrix [21,22,26] is defined as a set of N
rows, denoted Wj, for j = 0, 1, .... , N - 1, which have the
following properties:

         Wj takes on the values +1 and -1.
         Wj[0] = 1 for all j.                                                            1(b). „4-pattern‟ texture patterns
         WjxWkT=0, for j not equal to k and WjxWkT =N,
                                                                                       Figure 1. Generating 4 texture patterns
          for j=k.
         Wj has exactly j zero crossings, for j = 0, 1, .... , N-1.
         Each row Wj is even or odd with respect to its
          midpoint

Walsh transform matrix is defined using a Hadamard matrix of
order N. The Walsh transform matrix row is the row of the
Hadamard matrix specified by the Walsh code index, which
must be an integer in the range [0, ..., N - 1]. For the Walsh
code index equal to an integer j, the respective Hadamard
output code has exactly j zero crossings, for j = 0, 1, ... , N - 1.
     Using the Walsh-Hadamard transform assorted texture
patterns namely 4-pattern, 16-pattern and 64-pattern are
generated. To generate N2 texture patterns, each column of the
Walsh-Hadamard matrix of size NxN is multiplied with every
element of all possible columns of the same matrix (one
                                                                                    2(a). 4x4 Walsh-Hadamard transform matrix




                                                                       77                              http://sites.google.com/site/ijcsis/
                                                                                                       ISSN 1947-5500
                                                                (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                        Vol. 08, No.09, 2010
                                                                                                    1, if ....B(i, j )  TB
                                                                                   BMb(i, j )  {                                               (8)
                                                                                                 1,....if ...B(i, j )  TB

                                                                         To generate tiled bitmaps, the image is divided into four non-
                                                                         overlapping equal quadrants and the average of each quadrant
                                                                         is considered to generate the respective tile of the image
                                                                         bitmap.
                                                                                          VI.        DISCUSSED CBIR METHODS
                                                                         A. Mask-shape based CBIR
                                                                             In this method the feature vector is obtained by extracting
                                                                         the shape of the image by using gradient operators (Prewitt,
                                                                         Robert or Sobel) with slope magnitude method. Then the
                                                                         feature vectors are compared pixel by pixel using the
                                                                         Euclidian distance. The limitation of this method is that it is
         2(b). First four ‟16-pattern‟ texture patterns                  dependent on the size of the image. For this method, size of all
     Figure 2. Generating first four of 16 texture patterns              the images in the database should be same as query image.

                  V.         GRADIENT IMAGE BITMAPS                      B. CBIR with Mask-shape and BTC
   Image bitmaps of colour image are generated using three                   First of all the shape of the image is extracted by using
independent red (R), green (G) and blue (B) components of                three gradient operators with slope magnitude method. The
Prewitt/Robert/Sobel image obtained using slope magnitude                average of the obtained shape feature is calculated. The feature
method to calculate three different thresholds. Let                      vector is obtained by calculating the average of all those
X={R(i,j),G(i,j),B(i,j)} where i=1,2,….m and j=1,2,….,n; be              values which are greater than the average of the shape feature
an m×n slope magnitude gradient of color image in RGB                    and average of all those values which are less than or equal to
space. Let the thresholds be TR, TG and TB, which could be               the average of the shape feature. Here the size of the feature
computed as per the equations given below as 3, 4 & 5.                   vector is constant and is independent of size of the image.

                                                                         C. Proposed Shape Texture Pattern based CBIR
               1 m n
         TR         R(i, j)
              m * n i 1 j 1
                                                          (3)            In the proposed gradient shape texture method, the shape
                                                                         feature of the image is extracted using the three gradient
                                                                         operators Prewitt, Robert and Sobel. Then the bitmap of the
                                                                         shape feature is generated using the modified BTC technique.
                    1 m n
         TG              G(i, j)
                   m * n i 1 j 1
                                                          (4)
                                                                         The bitmap thus obtained is compared with the different
                                                                         texture patterns like „4-pattern‟, „16-pattern‟ and „64-pattern‟
                                                                         generated using Walsh-Hadamard transform matrix to produce
                                                                         the feature vector as the matching number of ones and minus
                    1 m n
                          B(i, j)
                                                                         ones per texture pattern. The size of the feature vector of the
         TB                                              (5)            image is given by equation 8.
                   m * n i 1 j 1
Here three binary bitmaps will be computed as BMr, BMg and
                                                                          Feature vector size=2*3*(no. of considered texture-pattern)                 (9)
BMb. If a pixel in each component (R, G, and B) is greater
than or equal to the respective threshold, the corresponding
pixel position of the bitmap will have a value of 1 otherwise it         Using three different gradient operators in association with
will have a value of -1.                                                 three assorted texture pattern sets along with original and
                                                                         original-even image, total 18 novel feature vector generation
                           1, if ....R(i, j )  TR                      methods can be used resulting into 18 new image retrieval
          BMr(i, j )  {
                                                                         techniques. Mask-shape based CBIR methods [30] are
                                                          (6)
                            1,....if ...R(i, j )  TR
                                                                         considered to compare the performance of proposed CBIR
                                                                         techniques. In the        proposed CBIR techniques           the
                                                                         combination of original and even part of images gives better
                           1, if ....G(i, j )  TG
                                                                         results than original image alone [1,2]. The main advantage of
          BMg(i, j )  {                                  (7)            proposed CBIR methods is improved performance resulting
                        1,....if ...G(i, j )  TG                       into better image retrieval. Here also the feature vector size is
                                                                         independent of image size in proposed CBIR methods.




                                                                    78                                       http://sites.google.com/site/ijcsis/
                                                                                                             ISSN 1947-5500
                                                                              (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                                      Vol. 08, No.09, 2010
                                                                                        category) are fired on the image database. The feature vector
     Table 1. Feature vector size of discussed image retrieval techniques
                                                                                        of query image and database image are matched using the
                                                                                        Euclidian distance. The average precision and recall values are
                          Feature                              Feature                  found for all the proposed CBIR methods. The intersection of
     CBIR                                   CBIR
                        vector size                          vector size                precision and recall values gives the crossover point. The
    Techniq                                Techniqu
                         for NxN                              for NxN                   crossover point of precision and recall is computed for all the
      ue                                      e
                          image                                image                    proposed CBIR methods. The one with higher value of
     Mask-                                                                              crossover point indicates better performance.
                            NxN            4-Pattern                  8
     shape
     Mask-
                                              16-                                                                      Prewitt       Robert        Sobel
    Shape +                   2                                   32




                                                                                           Crossover point of
                                                                                           Precision & Recall
                                            Pattern
     BTC                                                                                                        0.45
                                              64-                                                                0.4
                                                                 128                                            0.35
                                            Pattern
                                                                                                                 0.3
                                                                                                                0.25
                                                                                                                 0.2
                          VII.    IMPLEMENTATION
   The implementation of the discussed CBIR techniques is
done in MATLAB 7.0 using a computer with Intel Core 2 Duo
Processor T8100 (2.1GHz) and 2 GB RAM.
   The CBIR techniques are tested on the Wang image
database [18] of 1000 variable size images spread across 11                             Figure 3. Performance comparison of proposed CBIR methods with the mask-
                                                                                                                shape based CBIR methods
categories of human being, animals, natural scenery and
manmade things, etc. The categories and distribution of the                             Figure 3 shows the performance comparison of proposed
images is shown in table 2.                                                             CBIR methods with the mask-shape based CBIR methods
                                                                                        [30]. It is observed that the „mask-shape‟ based image retrieval
              Table 1. Image Database: Category-wise Distribution
                                                                                        gives the worst performance. However the „4-pattern‟ texture
  Category               Tribes              Buses              Beaches                 based image retrieval with Robert as gradient operator has the
    No. of                                                                              highest crossover point thus indicating better performance. It
                            85                 99                     99                is also observed that in case of „mask-shape‟ based image
   Images
  Category               Horses           Mountains            Airplanes                retrieval technique, Prewitt operator gives the best results
                                                                                        followed by Robert and Sobel. On the other hand Robert
    No. of                                                                              outperforms the other two operators in case of mask-shape
                            99                 61                     100
   Images                                                                               texture based image retrieval techniques.
  Category             Dinosaurs          Elephants              Roses
    No. of                                                                                                       Mask-Shape               Mask-Shape + BTC
                            99                 99                     99
   Images
                                                                                                                 4PatternTexture          16PatternTexture
                                                                                           Crossover point of
                                                                                           Precision & Recall




  Category           Monuments              Sunrise
    No. of                                                                                                       64PatternTexture
                            99                 61
   Images
                                                                                                                0.5
To assess the retrieval effectiveness, we have used the                                                         0.4
precision and recall as statistical comparison parameters [4,5]
for the proposed CBIR techniques. The standard definitions                                                      0.3
for these two measures are given by the equations 10 and 11.                                                    0.2
                 Number _ of _ relevant _ images _ retrieved                                                            Prewitt            Robert             Sobel
Pr ecision                                                                 (10)
                  Total _ number _ of _ images _ retrieved
                                                                                           Figure 4. Performance comparison of the proposed CBIR methods with
                  Number _ of _ relevant _ images _ retrieved
Re call                                                                    (11)                                different gradient operators
            Total _ number _ of _ relevent _ images _ in _ database
                                                                                        Figure 4 shows performance comparison of the proposed
                                                                                        CBIR methods with different gradient operators namely
                                                                                        Prewitt, Robert and Sobel. In case of Prewitt and Sobel, as the
                    VIII.    RESULTS AND DISCUSSION
                                                                                        number of texture patterns are increased (upto „16-pattern‟
  For testing the performance of each proposed CBIR                                     texture) the value of precision-recall crossover point also
method, 55 queries (randomly selected 5 from each image                                 increases. Beyond ‟16-pattern‟ texture, the results start



                                                                                   79                                            http://sites.google.com/site/ijcsis/
                                                                                                                                 ISSN 1947-5500
                                                                                             (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                                                     Vol. 08, No.09, 2010
deteriorating. In case of Robert, „4-Pattern‟ texture gives the                                       In all the three types of texture based image retrieval using
highest crossover point and on increasing the number of                                               different gradient operators, Robert outperforms the other two
texture patterns, the results start degrading.                                                        gradient operators Prewitt and Sobel.
                                                                                                      From comparison of the mask-shape based image retrieval
                                                                                                      techniques [30] with the proposed CBIR methods it is
                                                      4 Texture Pattern (Original)                    observed that the „4-pattern‟ texture based image retrieval with
                                                      4 Texture Pattern (Original+Even)               the combination of original with even image part using Robert
                                                      16 Texture Pattern (Original)                   gradient operator gives the best result. In case of Prewitt and
   Crossover point of Precision & Recall




                                                      16 Texture Pattern (Original+Even)              Sobel operators, increasing the number of texture patterns
                                                      64 Texture Pattern (Original)                   ameliorates the image retrieval performance upto a certain
                                                      64 Texture Pattern (Original+Even)              level (‟16-pattern‟ texture) beyond which the results start
                                           0.46                                                       deteriorating.
                                                                                                                                IX.     CONCLUSION
                                           0.44
                                                                                                      Better image retrieval techniques are proposed using shape
                                           0.42                                                       texture patterns generated with help of Walsh-Hadamard
                                                                                                      transform and gradient image bitmaps. Among the gradient
                                                                                                      operators “Robert” proved to be better for CBIR in proposed
                                            0.4
                                                                                                      methods. For Robert, „4-pattern‟ has given best performance
                                                                                                      and ‟16-pattern‟ is proven to be better in Sobel and Prewitt
                                           0.38                                                       operators. In all Robert 4-pattern has given best performance.
                                                     Prewitt       Robert         Sobel               The methods have been ameliorated using even image part
                                                                                                      along with original one for further improvement in CBIR
Figure 5. Amelioration of the proposed CBIR method with the                                           performance.
 combination of original and even image part using different                                                                     X.     REFERENCES
                     gradient operators
                                                                                                      [1]   Dr. H.B.Kekre, Sudeep D. Thepade, Varun K. Banura, “Augmentation
Figure 5 shows amelioration of the proposed CBIR methods                                                    of Colour Averaging Based Image Retrieval Techniques using Even
with the combination of original and even image part using                                                  part of Images and Amalgamation of feature vectors”, International
different gradient operators namely Prewitt, Robert and Sobel.                                              Journal of Engineering Science and Technology (IJEST), Volume 2,
                                                                                                            Issue 10, (ISSN: 0975-5462) Available online at http://www.ijest.info
It is observed that in case of Prewitt and Sobel operators, the
                                                                                                      [2]   Dr. H.B.Kekre, Sudeep D. Thepade, Varun K. Banura, “Amelioration
combination of original with even image part for „16-pattern‟                                               of Colour Averaging Based Image Retrieval Techniques using Even
texture gives the highest precision-recall crossover point. In                                              and Odd parts of Images”, International Journal of Engineering Science
case of Robert operator, the combination of original with even                                              and Technology (IJEST), Volume 2, Issue 9, (ISSN: 0975-5462)
image part for „4-pattern‟ texture gives the best performance.                                              Available online at http://www.ijest.info.
                                                                                                      [3]   Dr. H.B.Kekre, Sudeep D. Thepade, Akshay Maloo, “Query by Image
                                                                                                            Content Using Colour Averaging Techniques”, International Journal of
                                                                                                            Engineering Science and Technology (IJEST), Volume 2, Issue 6,
                                                                                                            2010.pp.1612-1622 (ISSN: 0975-5462) Available online at
                                             Prewitt Original        Prewitt Original+Even                  http://www.ijest.info.
                                                                                                      [4]   Dr. H.B.Kekre, Sudeep D. Thepade, “Boosting Block Truncation
                                             Sobel Original          Sobel Original+Even
   Crossover point of Precision




                                                                                                            Coding using Kekre‟s LUV Color Space for Image Retrieval”, WASET
                                                                                                            International Journal of Electrical, Computer and System Engineering
                                             Robert Original         Robert Original+Even                   (IJECSE), Volume 2, Number 3, pp. 172-180, Summer 2008. Available
                                                                                                            online at http://www.waset.org/ijecse/v2/v2-3-23.pdf
                                             0.46
                                                                                                            Dr. H.B.Kekre, Sudeep D. Thepade, “Image Retrieval using
            & Recall




                                                                                                      [5]
                                             0.44                                                           Augmented Block Truncation Coding Techniques”, ACM International
                                             0.42                                                           Conference on Advances in Computing, Communication and Control
                                              0.4                                                           (ICAC3-2009), pp. 384-390, 23-24 Jan 2009, Fr. Conceicao Rodrigous
                                             0.38                                                           College of Engg., Mumbai. Is uploaded on online ACM portal.
                                                                                                      [6]   Dr. H.B.Kekre, Sudeep D. Thepade, “Scaling Invariant Fusion
                                                      4 Texture     16 Texture    64 Texture                of Image Pieces in Panorama Making and Novel Image
                                                      Patterns       Patterns      Patterns                 Blending Technique”, International Journal on Imaging (IJI),
                                                                                                            www.ceser.res.in/iji.html, Volume 1, No. A08, pp. 31-46, Autumn
                                                                                                            2008.
 Figure 6. Performance comparison of the proposed CBIR methods using the
                 combination of original and even image part                                          [7]   Hirata K. and Kato T. “Query by visual example – content-based image
                                                                                                            retrieval”, In Proc. of Third International Conference on Extending
Figure 6 shows the performance comparison of the proposed                                                   Database Technology, EDBT‟92, 1992, pp 56-71
CBIR methods using the combination of original and even                                               [8]   Dr. H.B.Kekre, Sudeep D. Thepade, “Rendering Futuristic Image
image part. It is observed that the combination of original with                                            Retrieval System”, National Conference on Enhancements in Computer,
                                                                                                            Communication and Information Technology, EC2IT-2009, 20-21 Mar
even image part gives better performance than original alone.



                                                                                                 80                                   http://sites.google.com/site/ijcsis/
                                                                                                                                      ISSN 1947-5500
                                                                          (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                                  Vol. 08, No.09, 2010
       2009, K.J.Somaiya College of Engineering, Vidyavihar, Mumbai-77.                       A09,      Autumn      2009,pp.    55-65.    Available    online    at
[9]    Minh N. Do, Martin Vetterli, “Wavelet-Based Texture Retrieval Using                    www.ceser.res.in/iji.html (ISSN: 0974-0627).
       Generalized Gaussian Density and Kullback-Leibler Distance”, IEEE               [25]   Dr. H.B.Kekre, Sudeep D. Thepade, Akshay Maloo “Performance
       Transactions On Image Processing, Volume 11, Number 2, pp.146-158,                     Comparison for Face Recognition using PCA, DCT &WalshTransform
       February 2002.                                                                         of Row Mean and Column Mean”, ICGST International Journal on
[10]   B.G.Prasad, K.K. Biswas, and S. K. Gupta, “Region –based image                         Graphics, Vision and Image Processing (GVIP), Volume 10, Issue II,
       retrieval using integrated color, shape, and location index”,                          Jun.2010,              pp.9-18,           Available            online
       International Journal on Computer Vision and Image Understanding                       http://209.61.248.177/gvip/Volume10/Issue2/P1181012028.pdf..
       Special Issue: Colour for Image Indexing and Retrieval, Volume 94,              [26]   Dr. H.B.Kekre, Sudeep D. Thepade, “Improving the Performance of
       Issues 1-3, April-June 2004, pp.193-233.                                               Image Retrieval using Partial Coefficients of Transformed Image”,
[11]   Dr. H.B.Kekre, Sudeep D. Thepade, “Creating the Color Panoramic                        International Journal of Information Retrieval, Serials Publications,
       View using Medley of Grayscale and Color Partial Images ”, WASET                       Volume 2, Issue 1, 2009, pp. 72-79 (ISSN: 0974-6285)
       International Journal of Electrical, Computer and System Engineering            [27]   Dr. H.B.Kekre, Sudeep D. Thepade, Archana Athawale, Anant Shah,
       (IJECSE), Volume 2, No. 3, Summer 2008. Available online at                            Prathmesh Verlekar, Suraj Shirke, “Performance Evaluation of Image
       www.waset.org/ijecse/v2/v2-3-26.pdf.                                                   Retrieval using Energy Compaction and Image Tiling over DCT Row
[12]   Stian Edvardsen, “Classification of Images using color, CBIR Distance                  Mean and DCT Column Mean”, Springer-International Conference on
       Measures and Genetic Programming”, Ph.D. Thesis, Master of science                     Contours of Computing Technology (Thinkquest-2010), Babasaheb
       in Informatics, Norwegian university of science and Technology,                        Gawde Institute of Technology, Mumbai, 13-14 March 2010, The paper
       Department of computer and Information science, June 2006.                             will be uploaded on online Springerlink.
[13]   R.C. Gonsalez, R.E. Woods “Digital Image Processing”, Second                    [28]   Dr. H.B.Kekre, Tanuja K. Sarode, Sudeep D. Thepade, Vaishali
       Edition, Pearson Publication                                                           Suryavanshi,“Improved Texture Feature Based Image Retrieval using
                                                                                              Kekre‟s Fast Codebook Generation Algorithm”, Springer-International
[14]   Zhibin Pan, Kotani K., Ohmi T., “Enhanced fast encoding method for
                                                                                              Conference on Contours of Computing Technology (Thinkquest-2010),
       vector quantization by finding an optimally-ordered Walsh transform
                                                                                              Babasaheb Gawde Institute of Technology, Mumbai, 13-14 March
       kernel”, ICIP 2005, IEEE International Conference, Volume 1, pp I -
                                                                                              2010, The paper will be uploaded on online Springerlink.
       573-6, Sept. 2005.
                                                                                       [29]   Dr. H.B.Kekre, Tanuja K. Sarode, Sudeep D. Thepade, “Image
[15]   Dr. H.B.Kekre, Sudeep D. Thepade, “Improving „Color to Gray and
                                                                                              Retrieval by Kekre‟s Transform Applied on Each Row of Walsh
       Back‟ using Kekre‟s LUV Color Space”, IEEE International Advanced
                                                                                              Transformed VQ Codebook”, (Invited), ACM-International Conference
       Computing Conference 2009 (IACC‟09), Thapar University, Patiala,
                                                                                              and Workshop on Emerging Trends in Technology (ICWET
       INDIA, 6-7 March 2009. Is uploaded at online at IEEE Xplore.
                                                                                              2010),Thakur College of Engg. And Tech., Mumbai, 26-27 Feb 2010,
[16]   Dr. H.B.Kekre, Sudeep D. Thepade, “Image Blending in Vista Creation                    The paper is invited at ICWET 2010. Also will be uploaded on online
       using Kekre's LUV Color Space”, SPIT-IEEE Colloquium and                               ACM Portal.
       International Conference, Sardar Patel Institute of Technology, Andheri,
                                                                                       [30]   Dr. H.B.Kekre, Sudeep D. Thepade, Shobhit W., P. Mukharjee, S.
       Mumbai, 04-05 Feb 2008.
                                                                                              Singh, Miti K., “Image Retrieval with Shape Features Extracted using
[17]   Dr. H.B.Kekre, Sudeep D. Thepade, “Color Traits Transfer to                            Gradient Operators and Slope Magnitude Technique with BTC”,
       Grayscale Images”, In Proc.of IEEE First International Conference on                   International Journal of Computer Applications (IJCA), Volume 6,
       Emerging Trends in Engg. & Technology, (ICETET-08), G.H.Raisoni                        Number 8, pp.28-23, September 2010. Available online at
       COE, Nagpur, INDIA. Uploaded on online IEEE Xplore.                                    http://www.ijcaonline.org/volume6/number8/pxc3871430.pdf
[18]   http://wang.ist.psu.edu/docs/related/Image.orig (Last referred on 23
       Sept 2008)
                                                                                                                  AUTHORS PROFILE
[19]   Dr. H.B.Kekre, Sudeep D. Thepade, “Using YUV Color Space to Hoist
       the Performance of Block Truncation Coding for Image Retrieval”,
       IEEE International Advanced Computing Conference 2009 (IACC‟09),                                Dr. H. B. Kekre has received B.E. (Hons.) in Telecomm.
       Thapar University, Patiala, INDIA, 6-7 March 2009.                                              Engineering. from Jabalpur University in 1958, M.Tech
                                                                                                       (Industrial Electronics) from IIT Bombay in 1960,
[20]   Dr. H.B.Kekre, Sudeep D. Thepade, Archana Athawale, Anant Shah,                                 M.S.Engg. (Electrical Engg.) from University of Ottawa in
       Prathmesh Verlekar, Suraj Shirke,“Energy Compaction and Image                                   1965 and Ph.D. (System Identification) from IIT Bombay
       Splitting for Image Retrieval using Kekre Transform over Row and                                in 1970 He has worked as Faculty of Electrical Engg. and
       Column Feature Vectors”, International Journal of Computer Science                              then HOD Computer Science and Engg. at IIT Bombay. For
       and Network Security (IJCSNS),Volume:10, Number 1, January 2010,                                13 years he was working as a professor and head in the
       (ISSN: 1738-7906) Available at www.IJCSNS.org.                                                  Department of Computer Engg. at Thadomal Shahani
[21]   Dr. H.B.Kekre, Sudeep D. Thepade, Archana Athawale, Anant Shah,                                 Engineering. College, Mumbai. Now he is Senior Professor
       Prathmesh Verlekar, Suraj Shirke, “Walsh Transform over Row Mean                                at MPSTME, SVKM‟s NMIMS University. He has guided
       and Column Mean using Image Fragmentation and Energy Compaction                                 17 Ph.Ds, more than 100 M.E./M.Tech and several
       for Image Retrieval”, International Journal on Computer Science and                             B.E./B.Tech projects. His areas of interest are Digital Signal
       Engineering (IJCSE),Volume 2S, Issue1, January 2010, (ISSN: 0975–                               processing, Image Processing and Computer Networking. He
       3397). Available online at www.enggjournals.com/ijcse.                                          has more than 320 papers in National / International
[22]   Dr. H.B.Kekre, Sudeep D. Thepade, “Image Retrieval using Color-                                 Conferences and Journals to his credit. He was Senior
       Texture Features Extracted from Walshlet Pyramid”, ICGST                                        Member of IEEE. Presently He is Fellow of IETE and Life
       International Journal on Graphics, Vision and Image Processing                                  Member of ISTE Recently ten students working under his
       (GVIP), Volume 10, Issue I, Feb.2010, pp.9-18, Available online                                 guidance have received best paper awards and two have been
       www.icgst.com/gvip/Volume10/Issue1/P1150938876.html                                             conferred Ph.D. degree of SVKM‟s NMIMS University.
[23]   Dr. H.B.Kekre, Sudeep D. Thepade, “Color Based Image Retrieval                                  Currently 10 research scholars are pursuing Ph.D. program
                                                                                                       under his guidance.
       using Amendment Block Truncation Coding with YCbCr Color Space”,
       International Journal on Imaging (IJI), Volume 2, Number A09,
       Autumn 2009, pp. 2-14. Available online at www.ceser.res.in/iji.html.
[24]   Dr. H.B.Kekre, Tanuja Sarode, Sudeep D. Thepade, “Color-Texture
       Feature based Image Retrieval using DCT applied on Kekre‟s Median
       Codebook”, International Journal on Imaging (IJI), Volume 2, Number




                                                                                  81                                   http://sites.google.com/site/ijcsis/
                                                                                                                       ISSN 1947-5500
                                                        (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                Vol. 08, No.09, 2010
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, Mumbai. He has about than 07 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 has been
on International Advisory Board of many International
Conferences. He is Reviewer for many reputed International
Journals. 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 International Conference
SSPCCIN-2008, Second Best Paper Award at ThinkQuest-
2009 National Level paper presentation competition for
faculty and Best Paper Award at Springer International
Conference ICCCT-2010.


Varun K. Banura is currently pursuing B.Tech. (CE) from
MPSTME, NMIMS University, Mumbai. His areas of
interest are Image Processing and Computer Networks. He
has 05 research papers in International Conferences/Journals
to his credit.




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

								
To top