The matlab file contains two variables: fileanames - A cell array with names of all files images - A matrix of size num_files x 1000. with an IDF weight of each feature(1..1000) for each image file. market is a file with a matrix market format. See http://math.nist.gov/MatrixMarket/formats.html The file visterms_1000.market uses a dictionary of size 1K The file visterms_10k.market uses a dictionary of size 10K