Image Annotation via Reconstitution Graph Learning Model

Algorithm 1.

Framework of the proposed method.

Input: images

Output: predicted labels

1: Find the best nearest neighbor images by improving the nearest neighbors.

2: Construct a similar matrix through .

3: Mine the deep relationship between images, using random dot product graph (RDPG) for refactoring, .

4: Iterate to convergence through .

5: Build a semantic matrix through

6: Consider the effect of the association between labels on the results of the annotations, .

7: Consider the relationship between images and labels, .

8: Return the final score of the label, .

Algorithm 1.