MySQL Forums
Forum List  »  Newbie

appropriate database design and query
Posted by: gaurav mishra
Date: January 14, 2009 11:10AM

hi,
I am working on the project object recognition the thing is our in our algo we extract a very large sized feature vector from each image(ie 2000 values) representing
one vector.and we have about 40,000 images.


details of feature vector
in our algorithm we have 5 types of feature.there are 8 possible values for the first feature 5 for second and then 4 possible values for rest of features.
the vector which we store is that how many pixels belong to each of the possible combination of each type of feature there fore we have 8x5x4x4x4
columns in our vector.

i want to know what will be the best design for our database so that searching will be fastest.

currently we have split the database into 8 tables each corresponding to each type of first feature each table have 320 columns .....
the problem is querying the data is taking a very long time around 40 secs.......

we need to query for all the images which are closest to the query image....
closeness means all the columns values differ not more than 20 from input image feature values

Options: ReplyQuote


Subject
Written By
Posted
appropriate database design and query
January 14, 2009 11:10AM


Sorry, you can't reply to this topic. It has been closed.

Content reproduced on this site is the property of the respective copyright holders. It is not reviewed in advance by Oracle and does not necessarily represent the opinion of Oracle or any other party.