The results of the comparison are illustrated as graphs between Precision and Recall.
These quantitities are defined in literature as follows:-
PRECISION: Ratio of relevant images to total number of images retrieved in the query.Example: if 6 images are retrieved and only 3 belong to the category being searched we have 50% precision.
RECALL: Ratio of images retrieved in a query to total number of images in the database.Example if only the top 6 matches are retrieved from a database containing 60 images recal is 10 percent.
System performances are characterized by precision vs. recall graph which shows in general, how precision decreases as increasingly large fractions of
the collection are retrieved. An ideal Precision versus recall graph has precision = 1 for all values of recall (all relevant images are retrieved before any irrelevant ones).
The following graphs illustrate the retrieval performance on the 10 different
image categories each containing 100 images in terms of precision and recall.
The autocorrelogram performs better than the color histogram in 7 out of the 10
categories.
