Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Look up keyword
Like this
4Activity
0 of .
Results for:
No results containing your search query
P. 1
Content Base Image Retrieval Using Phong Shading

Content Base Image Retrieval Using Phong Shading

Ratings: (0)|Views: 56 |Likes:
Published by ijcsis
The digital image data is rapidly expanding in quantity and heterogeneity. The traditional information retrieval techniques does not meet the user’s demand, so there is need to develop an efficient system for content based image retrieval. Content based image retrieval means retrieval of images from database on the basis of visual features of image like as color, texture etc. In our proposed method feature are extracted after applying Phong shading on input image. Phong shading, flattering out the dull surfaces of the image The features are extracted using color, texture & edge density methods. Feature extracted values are used to find the similarity between input query image and the data base image. It can be measure by the Euclidean distance formula. The experimental result shows that the proposed approach has a better retrieval results with phong shading.
.
The digital image data is rapidly expanding in quantity and heterogeneity. The traditional information retrieval techniques does not meet the user’s demand, so there is need to develop an efficient system for content based image retrieval. Content based image retrieval means retrieval of images from database on the basis of visual features of image like as color, texture etc. In our proposed method feature are extracted after applying Phong shading on input image. Phong shading, flattering out the dull surfaces of the image The features are extracted using color, texture & edge density methods. Feature extracted values are used to find the similarity between input query image and the data base image. It can be measure by the Euclidean distance formula. The experimental result shows that the proposed approach has a better retrieval results with phong shading.
.

More info:

Published by: ijcsis on Jun 30, 2010
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

09/27/2010

pdf

text

original

 
Content Base Image Retrieval Using Phong Shading
 Abstract:-
The digital image data is rapidly expanding inquantity and heterogeneity. The traditional informationretrieval techniques does not meet the user’s demand, so there isneed to develop an efficient system for content based imageretrieval. Content based image retrieval means retrieval of images from database on the basis of visual features of imagelike as color, texture etc. In our proposed method feature areextracted after applying Phong shading on input image. Phongshading, flattering out the dull surfaces of the image Thefeatures are extracted using color, texture & edge densitymethods. Feature extracted values are used to find the similaritybetween input query image and the data base image. It can bemeasure by the Euclidean distance formula. The experimentalresult shows that the proposed approach has a better retrievalresults with phong shading.
 Keywords: -
 
CBIR, Gray scale, Feature vector, Phong shading,Color, Texture, Edge density.
I.
 
I
NTRODUCTION
 
In recent years, Content Based Image Retrieval (CBIR) hasplayed an important role in many fields, such as medicine,geography, security. General approaches in CBIR are basedon visual attributes of images such as color, texture, shapeetc. Most of CBIR systems are designed to find the top Nimages that are most similar to the input query image [2].The most common categories of descriptors are based oncolor, texture and shape. An efficient image retrievalsystem must be based on efficient image featuredescriptors. Image retrieval methods may also depend onthe properties of the images being analyzed [3]. Thepurpose of this paper is to develop a CBIR system usingthe concept of phong shading. The proposed solution is tofirst apply the phong shading on query image to enhancethe visual quality and then extract the visual features of query image and compare them to the database featuresused are color texture & edge density. The mean, medianand standard deviation of red, green and blue channels of color histogram. The texture features such as contrast,energy & entropy are retrieved from phong shaded grayscale image. The edge feature includes vertical andhorizontal edges Euclidean distance formulas have beenused for similarity measurement [1].II P
ROPOSED FRAMEWORK
 
Proposed frame work is as follows:-In proposed framework first apply the phong shading oneach image of database then extract its color texture & edgefeatures and then construct single feature vector for eachimage. Feature database has feature vector of each imageand same procedure is applied on query image to extract itsfeatures then compare its feature vector with featuredatabase to retrieve the most similar images.II.
 
P
HONG
 
SHADING
The interpolation method may also be called Phong
 
interpolation, which is usually referred to by "per-pixellighting". Typically it is called "shading".An illumination model define as model for determining aradiant intensity at a given point in space, for single lightsource it can be formulated as follows:
()()
S
 N a a d l s l
 I K I K I N L K I N
= + • +
  
(1)
SANJEEV JAINProfessor & Head of CSE Department
LNCT, Bhopal (M.P) INDIA
dr_ sanjeevjain@yahoo.com
 
UDAY PRATAP SINGH,Assistant Professor of CSE Department.
LNCT, Bhopal (M.P) INDIA
usinghiitg@gmail.com
 
GULFISHAN FIRDOSE AHMED,M.Tech. Student of CSE Department.
LNCT, Bhopal (M.P) INDIA
gul_firdose@rediffmail.com
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 8, No. 1, April 2010301http://sites.google.com/site/ijcsis/ISSN 1947-5500
 
This formula states that the intensity at a certain pointdepends on the reflectance of the object (
a
,
and
s
are ambient, diffuse, and specular reflectance,respectively), on the intensities of the light source
l
 I 
, andon some vectors.
  
denotes the unit normal vector to thesurface.
 L
  
denotes the unit light source vector, and
 H 
uur
isthe unit halfway vector indicating the direction of themaximum highlight. Another object dependent feature isexpressed by the glossiness
S
 N 
which ranges from one toinfinity. Phong shading is a well known method forproducing realistic shading but it has not been used by real-time systems The phong shading is basically used toenhance the visual quality of an image to extract featuresmore accurately. The Phong model describes theinteraction of light with a surface, in terms of the propertiesof the surface and the nature of the incident light. ThePhong model reflected light in terms of a diffuse andspecular component together with an ambient term. Theintensity of a point on a surface is taken to be the linearcombination of these three components. Phong shadingcalculates the average unit normal vector at each of thepolygon vertices and then interpolates the vertex normalover the surface of the polygon after calculating the surfacenormal it applies the illumination model to get the colorintensity for each pixel of the polygon.
Figure2-
 
Phong shading (interpolation of vectors)
The intensity of light on the surface is the combination of diffuse reflection, ambient light & specular reflection theintensity of diffused light is given in fig.2For phong shading
 N
   
of the specular term can beformulated as
nnnnnnnn
 N  N  N
=
       
 
( ) ( )( ) ( )
 Ax By C Dx Ey  Ax By C Dx Ey
+ + + +=+ + + +
    
(2)where
nn
 N Ax By
= + +
  
represents the approximatenonunit length normal vector that is linearly interpolatedfrom the true surface normal vectors specified at thevertices.
,,
 A B C 
  
are chosen to interpolate the surfacenormal vector across the polygon.
nn
 H Dx Ey
= + +
  
 represents the nonunit length halfway vector between thelight source vector and viewer vector.
 
III.
 
C
OLOR FEATURES
 
One of the most important features that make possible therecognition of images by humans is a color. Color is aproperty that depends on the reflection of light to the eyeand the processing of that information in the brain. Themain method of representing color information of imagesin CBIR systems is through color histograms. Forextracting the color features, the color histogram is used asit is independent of image size & orientation. Here theRGB color histogram is used. for extracting these featurefirst we find the histogram for red, green & bluechannel..[1, 3, 4, 5, 10]IV.
 
T
EXTURE
 
FEATURES
 
Texture is that innate property of all surfaces that describesvisual patterns, each having properties of homogeneity. Itcontains important information about the structuralarrangement of the surface, such as; clouds, leaves, bricks,fabric, etc. The texture features describes the relationshipof the surface to the surrounding environment. A texture ischaracterized by a set of values called energy, entropy,contrast, and homogeneity.In short, it is a feature thatdescribes the distinctive physical composition of asurface.The following formulas are used to calculate thefeatures and are shown in equations 3 to 6 [1, 9]:-
 A.
 
 Entropy: -.
 Entropy is defined as
( ) ( )
,log,
i j
 Entropy P i j P i j
=
(3)Where P contains the histogram counts.
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 8, No. 1, April 2010302http://sites.google.com/site/ijcsis/ISSN 1947-5500
 
 B.
 
Contrast:-
It is the measure of the intensity contrast between apiel and its neighbor over the whole image.
( ) ( )
2
,
i j
Contrast i j P i j
=
 
(4)
C.
 
. Energy: -
 It is the sum of squared elements.
( )
2
,
i j
 Energy P i j
=
 (5) 
 D.
 
 Homogeneity: -
 A value that measures the closeness of the distributionof elements . 
( )
,1
i j
P i j Homogeneityi j
=+
(6)V.
 
E
DGE DETECTION
 
Edge detection refers to the process of identifying andlocating sharp discontinuities in an image. Thediscontinuities are abrupt changes in pixel intensity whichcharacterize boundaries of objects in a scene. Edgedetection is a terminology in image processing andcomputer vision, particularly in the areas of featuredetection and feature extraction, to refer to algorithmswhich aim at identifying points in a digital image at whichthe image brightness changes sharply or more formally hasdiscontinuities. Edge detection would then be done on theintensity channel of a color image in HSI space. Anotherdefinition claims an edge exists if it is present in the redgreen and the blue channel. Performing it on each of thecolor component can do edge detection.Edge detectionrefers to the process of identifying and locating sharpdiscontinuities in an image. The discontinuities are abruptchanges in pixel intensity which characterize boundaries of objects in a scene. [1, 10]
 A.
 
Sobel Edge detection:-
The Sobel operator is used in image processing,particularly within edge detection algorithms. The Sobeloperator is based on convolving the image with a small,separable, and integer valued filter in horizontal andvertical direction and is therefore relatively inexpensive interms of computations. The operator consists of a pair of 3×3 convolution kernels mask as shown below. One kernelis simply the other rotated by 90°.-1 0 +1 +1 +2 +1-2 0 +2 0 0 0-1 0 +1 -1 -2 -1Gx GyVI.
 
S
IMILARITY COMPARISON
 
CBIR employs low level image features such as color,shape or texture to achieve objective and automaticindexing, in contrast to subjective and manual indexing intraditional image indexing. For contend
based image
retrieval, the image feature extracted is usually an N-dimensional feature vector which can be regarded as apoint in a N-dimensional space. A similarity measurementis normally defined as a metric distance.For similaritycomparison we have used the Euclidean distance formula.Two pixels has coordinates (X
1
, Y
1
) and (X
2
, Y
2
) then theEuclidean distance is given by:
( ) ( )
222121
 Euclid 
 D X X Y
= +
(7)VII.
 
E
XPERIMENTAL RESULT AND ANALYSIS
 
For retrieval efficiency and effect of phong shading on animage. From the corel database 70 images with fivedifferent subjects each contain 14 images are taken.Calculate a recall & precision value in both cases outputwith phong shading and without phong shading on animage. Recall and precision depend on the outcome of aquery and all its relevant and non-relevant images.Standard formulas have been used to compute the precisionand recall for query images is shown below.Precision is defined as the number of relevantdocuments retrieved by a search divided by the totalnumber of documents retrieved by that search.Recall is defined as the number of relevantdocuments retrieved by a search divided by the totalnumber of existing relevant documents (which should havebeen retrieved).For calculating recall & precision value we have considerdatabase of 70 images randomly selected from Caltech_256image database. In our database we have consider 5different categories and each categories contain 14 imageeach.Fig3 (without phong shaded) and Fig4(with phong shaded)is shown below. Feature value becomes are same but in
..
 Noof releventimagesretrieved  precisionTotalnoof imagesretrieve
=
 
No of relevant images retrievedTotal no. of relevant images in the database
 Recall
=
 
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 8, No. 1, April 2010303http://sites.google.com/site/ijcsis/ISSN 1947-5500

You're Reading a Free Preview

Download
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->