The effects of surrounding text in image indexing

Indexing images on the web has become increasingly challenging. Numerous researches have been conducted to improve the indexing of the images. This research looks into manipulating the surrounding text factor in an effort to increase the success rate of image retrieval. A series of test involving a total of 5, 10, 20 and 30 text before and after the images was conducted to observe the effects on the precision-recall value. The resulting index was then compared to the level of image retrieval relevance. From the analysis, it was found that if the number of surrounding text taken into consideration is very small, it will not give much meaning to the image. If the number of surrounding text is large, it would also not give significant meaning to the image, but would give a negative effect to the precision-recall value. It is hoped that the findings could shed more light on the use of surrounding text factor in image indexing.