International Conference on Advances In Engineering And Technology - ICAET 2014
Author(s) : BHUMIKA SHAH, DHATRI PANDYA
With the detonative growth of the digital technologies in the web large amount of visual data are created and stored. The majority of image available on the web have little or no metadata associated with it describing the semantic concept associated with the images. There is a need of efficient and effective technique to find visual information on demand.One of the promising approach to enhance the image retrieval is automatic image annotation which refers to process of assigning relevant keywords to the image to bridge the semantic gap between low level content features of image such as color, texture and shape and semantic concepts understand by the humans such as keywords, description or image classification. The paper discusses implementation of the automatic image annotation using fuzzy c means clustering to annotate the image based on Dominant color and Gray level cooccurence matrix texture feature. The experiments are conducted on 50 beach images and 50 images of the corel dataset to identify the best cluster number and similarity measure such as Cityblock,Sqeculidean,Canberra used in the fuzzy c means clustering for annotating the image.The experiments results and comparison with other clustering method and feature descriptor showed that the proposed automatic image annotation scheme can effectively improve the annotation performance .The analysis of annotation results using various similarity measure and feature descriptor are also presented in the paper.