Notice of Violation of IEEE Publication PrinciplesBag-of-Features Based Medical Image Retrieval via Multiple Assignment and Visual Words Weighting

Jingyan Wang, Yongping Li, Ying Zhang, Chao Wang, Honglan Xie, Guoling Chen, Xin Gao
2011 IEEE Transactions on Medical Imaging  
In this paper, we investigate the bag-of-feature based medical image retrieval methods, which represent an image as a collection of local features, such as image patch and key points with SIFT descriptor. To improve bag-of-feature method, we first model the assignment of local descriptor as contribution functions, and then propose a new multiple assignment strategy. Assuming the local feature can be reconstructed by its neighboring visual words in vocabulary, we solve the reconstruction weights
more » ... using QP problem and use them as contribution functions, resulting a new assignment methods, called QP assignment. At the same time, we also propose a novel visual weighting method. We first analysis each visual word by modeling the subsimilarity or sub-distance function comparing only one single bin corresponding to the visual word; then we treat each of them as a weak classifier for triplets and learn a strong classifier, the resulting weights will be used as visual weighting factors. We carry our experiments on three medical image datasets: the ImageCLEFmed dataset, the 304 CT Set and the Basal-Cell Carcinoma Image set. The vast experiments results show that our proposed methods have many advantages and work well for the bag-of-feature based medical image retrieval tasks.
doi:10.1109/tmi.2011.2161673 pmid:21859616 fatcat:iyz5j26sdrdbxc4mvpe6b4ekn4