Content Based Medical Image Retrieval Using Multi-Trend Structure Descriptor and Fuzzy k-NN Classifier

In this work, Multi-trend structure descriptor is suggested for retrieving images from heterogeneous medical image dataset. The multi-trend structure descriptors acquires structures in the form of large, small and equal trends along four directions at both global and local level. For the retrieval of grayscale medical images, only the structure information computed by Multi-trend structure descriptor from the texture and edge information of an image is used while for color medical images the color information acquired by Multi-trend structure descriptor is also taken in to consideration for attaining effective retrieval rate. The fuzzy k-NN approach is incorporated to classify the feature vectors and it enhances the retrieval accuracy as well it increases the speed of retrieval. The results obviously shown that the fusion of Multi-trend structure descriptor and fuzzy k-NN outperforms the state-of-the-art approaches.