Builds a spatial pyramid matching SVM classifier

Description

Builds a spatial pyramid matching SVM classifier with convolutional descriptors that are vector quantized with sparse coding.

PYRAMID_LEVELS               the number of levels to form when building the spatial pyramid matching.
gridSpacing                            the spacing for the grid to be used when generating the sift descriptors
patchSize                               the patch size used for generating the sift descriptor
poolSize                                 the amount to pool the image patches down before collecting them on the grid of pooled pixels.
gridSize                                  the size of the regions on the pooled maps to collect x the number of pooled maps gives the size of the descriptor.
cropSize                                 the size of the crop region to take out of the middle of each feature map.
dictionarySize                         the size of the divtionary used in K-means.
temp_dir                                 temporary directory to save intermediate files
raid_dir                                   directory to save the results to.
train_image_dir                      where the texture images are stored for the train set.
test_image_dir                       where the texture images are stored for the test set.
end_data_dir                          where to save to.
train_image_names               the directory where each category has a folder (for getting the names of the training set).
test_image_names                the directory where each category has a folder (for getting the names of the testing set).

A linear subspace learning approach via sparse coding

Bayesian Structured Sparse Coding into Image Restoration