Inverted-File-Structure

A function to Create an Inverted File Structure for an image database for large scale retrieval

This function takes the SIFT like local features from an image database (features grouped imagewise) and a codebook(dictionary) to return an inverted file structure. i.e, the SIFT features of an image are mapped(quantized) to the nearest codewords in the dictionary. And each codeword holds the images(ids) containing features that are mapped to it. This will be a sparse list when the number oo codewords present in the dictionary is huge. This function calls one more functino(insert_image.m) to insert the image in the appropriate locations in the structure. One final note is, to use this function one must install the yael_library from the INRIA group, 'https://gforge.inria.fr/projects/yael/'. As a byproduct this function gives the un-normalised BoW(Bag of Visual Words) representation for the images in the database.