This paper presents an efficient content based image retrieval scheme for both the shaped and unshaped objects. The local regions of an unshaped image have been classified with respect to the frequency of occurrence. Then the semantic concept is evaluated through RGB histogram dissimilarity factor, overall dissimilarity factor and regional dissimilarity factor. These dissimilarities cooperatively determine the local concept for the unshaped object. In addition, the semantic concept for shaped objects is measured through the normalized color findings, synchronized edge detection, small unnecessary particle remotion, and shape similarity checking. All these measurements mutually rank the shaped objects according to their probability of occurrences. In addition, several algorithms and theoretical explanations of the proposed semantic models have been presented. The corresponding examples and simulations prove that the proposed methods work accurately. The comparative results show that the proposed models have significantly better scalability than the existing approaches.
|Number of pages||18|
|Journal||IOSR Journal of Computer Engineering|
|Publication status||Published - 2016|