06-10-2016, 12:59 PM
1458031021-confer.odt (Size: 1.49 MB / Downloads: 5)
Abstract:-This paper uses a technique which is named as CBIR (Content Based Image Retrieval).In this technique,we uses three fused features like color ,texture and shape. First extract the features from each module,then we perform extraction for fused features. Who image property can be achieved with the help of fused feature extraction. In color feature extraction, we use color histogram representation .Color histogram means it represents the image not in the same perspective. It includes two color histograms like global and local .First we convert RGB to HSV color space quantification for meeting the visual requirements. Then calculate the co- occurence matrix for the purpose of texture feature extraction. Finally gradient method is used to extract the shape features from the query image. Then the result will be shown for fused features .By using the fused features in CBIR, efficiency or enhancement of image retrieval is improved than the single feature extraction.
Keywords— Global color histogram, local color histogram, RGB,HSV, co-occurrence matrix, gradient method.
I. INTRODUCTION
Image retrieval is the photographs process of retrieve the images from the large database. Images may be represented as, hand sketches or query images. The most common image retrieval systems are text based image retrieval(TBIR) systems, where the search is based on automatic or
manual annotation of images. A conventional TBIR searches the database for the similar text surrounding the image as given in the query string. The commonly used TBIR system is Google Images. However, it is sometimes difficult to express the whole visual content of images in words and TBIR may end up in producing irrelevant results. In addition annotation of images is not always correct and consumes a lot of time. For finding the alternative way of searching and overcoming the limitations imposed byTBIR systems more intuitive and user friendly content based image systems (CBIR) were developed. High retrieval efficiency and less computational complexity are thedesired characteristics of CBIR systems retrieval.
Content-based image retrieval (CBIR) [1] is a technique which uses visual contents such as color, shape and texture for searching similar images from large scale image database according to user request in the form of query image. Color, texture and shape features have been used for describing image content.
Content-based image retrieval uses the visual contents of an image such as color , texture, shape to represent and index the image. In typical content-based image retrieval systems shown in figure1, the visual contents of the images in the database are extracted and described by multi-dimensional feature vectors. The feature vectors ofthe images in the database form a feature database. To retrieve images, users provide the retrieval system with query image. The system then changes these examples into its internal representation of feature vectors. The similarities distances between the feature vectors of the query image and those of the images in the database are then calculated and retrieval is performed with the aid of an indexing scheme. The indexing scheme provides an efficient way to search for the image database.
II .RELATED WORK
There are many approaches for Content Based Image Retrieval using different features such as color, texture and shape. Some of the related works are described below.
CBIR using fused features of color and texture which proposed an method to compare the two color features[1]. From the results, it is obviously to see that the global color histogram counts the overall image color information without spatial information. So a great characteristic is rotational invariance. Then we extract texture features with the help of co-occurence matrix.
Chin-Chin Lai et.al.[2] have proposed an interactive genetic algorithm (IGA) to reduce the gap between the retrieval results and the users’ expectation called semantic gap. They have used HSV color space that corresponds to human way of perceiving the colors and separate the luminance component from chrominance ones. They have also used texture features like the entropy based on the grey level co-occurrence matrix and the edge histogram. They compared this method with others approaches and achieved better results
A. Kannan et.al.[3]have proposed Clustering and Image Mining Technique for fast retrieval of Images. The main objective of the image mining is to remove the data loss and extracting the meaningful information to the human expected needs. The images are clustered based on RGB Components,Texture values and Fuzzy C mean algorithm.
RishavChakravarti et.al [4] have published a paper on Color Histogram Based Image Retrieval. They have used color histogram technique to retrieve the images. This method allows retrieval of images that have been transformed in terms of their size as well as translated through rotations and flips.
Ramesh Kumar et.al[5] have published on Content Based Image Retrieval using Color Histogram. They have used Color Histogram technique to retrieve the similar images. To speed up the retrieval, they have used the proposed grid-based indexing to obtain the nearest neighbours of the query image and exact images are retrieved. Indexing can be performed in vector space to improve retrieval speed. Mainly, they have implemented CBIR using color histogram technique and is refined with help of grid technique to improve the image retrieval performance.
III. SYSTEM OVERVIEW AND PROPOSED METHOD
3.1 Color Feature Extraction:
Color feature is one of the most widely used features in low level feature. Compared with shape feature and texture feature, color feature shows better stability and is more insensitive to the rotation and zoom of image. Color not only adds beauty to objects but also more information, which is used as powerful tool in content-based image retrieval
HSV color space is a popular choice for manipulating color. The HSV color space is developed to provide an intuitive representation of color and to approximate the way in which humans perceive and manipulate color. RGB to HSV is a nonlinear, but reversible, transformation. The hue (H) represents the dominant spectral component-color in its pure form, as in green, red, or yellow. Adding white to the pure colour changes the color: the less white,the more saturated the colour is. This corresponds to the saturation (S). The value (V) corresponds to the brightness of color. The coordinate system is cylindrical, and is often represented by a subspace defined by a six-sided inverted pyramid. The top of the pyramid corresponds to V=1, with the ―white‖ at the centre. The hue is measured by the angle around the vertical axis, with red corresponding to 0. The saturation ranges from 0 at the centre to 1 on the surface of the pyramid. An inverted cone is also used to denote the subspace instead of the pyramid.
The following steps are followed to extract color feature.
1. Read the query image from user.
2. Convert RGB colour space into HSV color space.
3. Quantize each pixel in HSV space to 256histogrambins.
4. The normalized histogram is obtained by dividing With the total number of pixels.
5. Store the 256 values as color feature vector in feature vector database.
6. Calculate the similarity measure of query image and the image present in the database using
Canberra Distance.
7. Retrieve the images based on minimum distance
3.2 Texture Feature Extraction
Grey Level Co-occurrence Matrix (GLCM) is a widely used texture descriptor and it is proven that results obtained from the co-occurrence matrix are better than the other texture discriminations methods . GLCM computes the statistical features based on grey level intensities of the image. It enhances the details of image and gives the interpretation. The GLCM is a tabulation of how often different combinations of pixel brightness values (gray levels) occur in an image. The advantage of the co occurrence matrix calculations is that the co-occurring pairs of pixels can be spatially related in various orientations with reference to distance and angular spatial relationships, as on considering the relationship between two pixels at a time. As a result the combination of grey levels and their positions are exhibited apparently. Therefore it is defined as ―A two dimensional histogram of grey levels for pair of pixels, which are separated by a fixed spatial relationship.
The following steps are followed to extract texture feature.
1. Read the query image from user.
2. Convert RGB query image to Grey Scale image.
3. Compute four GLCM matrices for each direction
4. For each GLCM matrix compute the statistical features such as Energy, Homogeneity, Contrast and Correlation.
5. Compare similarity matching of database image with query image using distance metrics.
6. Retrieve the top ten images based on minimum distance.
3.3 Shape Feature Extraction
Shape is the most important feature for recognizing objects. An image is a function of two variables f(x, y). If image is assigned value from 0 to 1 according to brightness of the image then for white color, pixel is assigned a zero value. For black color of pixel value1 is assigned and for grayish color, value between 0 and 1 is assigned depending on the brightness of that pixel. Rapid change in color intensity i.e. the sharp contrast indicates the edge in an image. A rapid change in a function gives a large magnitude of the gradient at edges. The gradient is the geometric computing method for characterizing symmetric breaking of an ensemble of asymmetric vectors regularly distributed in a square lattice. In gradient method, edges are detected first. It is done by looking for the maximum and minimum in the first derivative of the image.
The following steps are followed to extract Shape feature.
1. Read the query image from user.
2. Convert RGB query image to Grey Scale image.
3. Calculate 4 morphological gradients of edge maps are generated.
4. Calculate seven moment invariants for each edge map,totally28 features are stored.
5. Compare similarity matching with database image with query image using distance metrics.
6. Retrieve the top ten images based on minimum distance.
VI.COMBINING FEATURES
The retrieval result using only single feature may be inefficient. It may either retrieve images not similar to query image or may fail to retrieve images similar to query image. Hence,to produce efficient results, we use combination of color, shape and texture features. The similarity between query and target image is measured from three types of characteristic features which includes color, shape and texture features. So, during similarity measure, appropriate weights are considered to combine the features. The distance between the query image and the image in the database is calculated as follows:
CONCLUSION
Content based image retrieval has overcome all the limitations of text based image retrieval by considering the visual contents of image such as color, shape and texture. Here we successfully implemented the CBIR system by combining three features i.e., color, shape and texture. In this paper a new algorithm content based image retrieval is presented.(i) HSV color space.(ii)Edge detection and moment invariant.(iii)Grey Level Co-occurrence Matrix(GLCM) are combined to build a feature vector database. The proposed method gives better retrieval results and average precision. Wehave used only four GLCM statistical features with angle and distance d=1.Future work of the study are Grey level Co-occurrence matrices with different angle with different distances and in HSV color space different levels of H,S,V to generate histogram bins.