A Semisupervised Framework for Automatic Image Annotation Based on Graph Embedding and Multiview Nonnegative Matrix Factorization

Algorithm 1

Multiview NMF with graph embedding for image annotation.

Input: Image set and label matrix of the training set.

Output: Predicted label matrix for the test set .

(1) Extract different feature for image set ;

(2) Construct Laplacian graph and ;

(3) Initialize and randomly;

(4) do

(5) Update and based on equation (23)-(25);

(6) while the terminating condition is not satisfied

(7) Input into 2PKNN [21] image annotation algorithm;

(8) Output predicted labels for the test set.