Range Distance in Query-by-Example(s): An Application

S. Koompairojn, K.A. Hua, N. Hiransakolwong, and K. Vu (USA)


an application for CBIR, dynamic metric, range distance, content-based image retrieval, query-by example


For more than a decade, query-by-one-example has been a popular query system for content-based image retrieval (CBIR). All N dimensions of features (the static) have been used in the distance metric. The static distance metric measures the similarity between images. We propose a novel modified system using the range distances automatically adjust weights. This system modifies the static distance metric in to the dynamic distance metric for query-by-example(s). For our system, users can query using more than one query image or using more than one group of query images. Users can specify positive, negative or neutral groups. For each feature within these groups of query images, their range distances are used for adjusting weights. Therefore some features may not be employed in the distance metric. This distance metric becomes a dynamic distance metric. With our approach, we are able to achieve a higher degree of precision and recall, and at the same time, significantly reduce time complexity of matching. We tested our approach against the ImageGrouper method. The results show that our approach is an effective and efficient technique for query-by-example(s).

Important Links:

Go Back