To search, Click below search items.


All Published Papers Search Service


Image Retrieval Using Non-Binary-Weighted Approach


Khor Siak Wang, Fatimah Ahmad


Vol. 9  No. 9  pp. 101-106


Colour histogram has been a popular technique for colour indexing and image retrieval. However, it suffers from a major drawback, i.e. fails to take into consideration the spatial information of an image during the image matching process. In this paper, we present a new technique for retrieving images having similar chromatic content in a query image. This is accomplished by assigning non-binary weights to index terms in the query and stored images, a popular technique making use of the vector model from the literatures of information retrieval for image similarity match. The index terms are derived from the image content. The proposed technique will be benchmarked against the histogram technique. A standard dataset that currently contains 1338 colour images, known as Uncompressed Colour Image Database (UCID), is used for the benchmarking purpose. With the system developed using both Visual Basic and Java, it has shown an improved performance in term of retrieval accuracy, with an averaged precision value of about 70 % over the traditional histogram technique that only enjoyed approximately 20 % of precision value for retrieving images having similar chromatic content with a query image.


Vector Model, Hue Pattern, Bucket, Boolean Model, Vector Model, Probabilistic Model