Efficient Content Based Image Retrieval Using Combination Of Dominant-Color, Shape And Texture Features And K-Means Clustering

There is a huge demand for the efficient content based image retrieval system because of the

availability of large image databases. In this paper we have present an efficient CBIR framework by

extracting the Dominant-color, Texture, edge features and by clustering feature database. We have applied

the dominant color extraction using color-quantization technique. Initially the image is divided into some

partitions using the color quantization algorithm, here we are dividing into eight partitions and the eight

dominant colors are obtained from that partition. Next for shape feature extraction sobel color edge

detection technique is used. And local binary pattern (LBP) is performed on gray scale image to extract the

texture feature. Then all features discussed above of image are combined to form a single feature vector. Kmeans

clustering is applied over combined feature vector of database images. Finally, to retrieve similar

images from database similarity matching is performed by Euclidian distance which compares feature vector

of clustered database images with that of query image. The result of this proposed approach provides

efficient, more accurate result.