No abstract

No abstract

No abstract

No abstract

Recent works on Content Based Image Retrieval rely on bag of visual words to index visual content. Analogically to the bag of words approach in text retrieval, this model of description represents an image as a vector of weights, where each weight corresponds to the importance of a visual word in the image, and is computed according to the chosen weighting scheme. Instead of using the known weighting schemes directly migrated from text retrieval domain, we propose a new approach specifically for images. The proposed weighting scheme is based on a fuzzy model to take into account the fundamental difference that exists between textual words and visual words. For experiments, two datasets with very different properties are used. The tests clearly demonstrate that retrieval based on the proposed technique produces better results than standard term weighting schemes.