Open AccessIMAGE SEARCH ENGINE BASED ON COMBINED FEATURES OF IMAGE SUB-BLOCKS
Author(s) -
Chaithanya Sagar Kotapuri,
Mohammad Hayath Rivjee
Publication year - 2012
Publication title -
international journal of image processing and vision sciences
Language(s) - English
Resource type - Journals
ISSN - 2278-1110
DOI - 10.47893/ijipvs.2012.1018
Subject(s) - histogram , artificial intelligence , image texture , pattern recognition (psychology) , mathematics , histogram matching , image histogram , computer vision , image retrieval , block (permutation group theory) , computer science , image processing , image (mathematics) , combinatorics
In this paper we propose a new and efficient technique to retrieve images based on sum of the values of Local Histogram and GLCM (Gray Level Co-occurrence Matrix) texture of image sub-blocks to enhance the retrieval performance. The image is divided into sub blocks of equal size. Then the color and texture features of each sub-block are computed. Most of the image retrieval techniques used Histograms for indexing. Histograms describe global intensity distribution. They are very easy to compute and are insensitive to small changes in object translations and rotations. Our main focus is on separation of the image bins (histogram value divisions by frequency) followed by calculating the sum of values, and using them as image local features. At first, the histogram is calculated for an image sub-block. After that, it is subdivided into 16 equal bins and the sum of local values is calculated and stored. Similarly the texture features are extracted based on GLCM. The four statistic features of GLCM i.e. entropy, energy, inverse difference and contrast are used as texture features. These four features are computed in four directions (00, 450, 900, and 1350). A total of 16 texture values are computed per an image sub-block. An integrated matching scheme based on Most Similar Highest Priority (MSHP) principle is used to compare the query and target image. The adjacency matrix of a bipartite graph is formed using the sub-blocks of query and target image. This matrix is used for matching the images. Sum of the differences between each bin of the query and target image histogram is used as a distance measure for Local Histogram and Euclidean distance is adopted for texture features. Weighted combined distance is used in retrieving the images. The experimental results show that the proposed method has achieved highest retrieval performance.