how to find precision and recall for a Image retrieval system?

I have Image retrieval system , i have extracted the features, and testing , training is performed by Multisvm, I obtained Confusion matrix for that... but how to obtain the precision and recall plz help.

[hp]

matlab2013

Answer 1

[Image Analyst]

You have the confusion matrix so that gives you everything you need. What more do you want? How are you defining precision? Is it (# true positives) / (number of images)? Well, you can get that, plus more, from the confusion matrix. Please explain why the confusion matrix is not giving you everything you want.

[hp]

yeah it will give but simply i want that also... can you tell me how to draw ROC curve(for matlab2013a verson) for more than one classes if the classifier used is multi-class svm. as i came to know from the answers that we should consider binary class,keeping first class as one remaining as zero but how to do it Iam not getting plz help...