Graph Regularized Nonnegative Matrix Factorization with Sparse Coding

<table class="figure-group"><tr class="fig-image" id="a"><td><object data="https://static.hindawi.com/articles/mpe/volume-2015/239589/figures/239589.fig.005a.svgz" name="239589.fig.005a" type="image/svg+xml"></object></td></tr><tr class="fig-caption"><td><b>(a) </b>Occluding patch size 10 × 10</td></tr><tr class="fig-image" id="b"><td><object data="https://static.hindawi.com/articles/mpe/volume-2015/239589/figures/239589.fig.005b.svgz" name="239589.fig.005b" type="image/svg+xml"></object></td></tr><tr class="fig-caption"><td><b>(b) </b>Occluding patch size 15 × 15</td></tr><tr class="fig-image" id="c"><td><object data="https://static.hindawi.com/articles/mpe/volume-2015/239589/figures/239589.fig.005c.svgz" name="239589.fig.005c" type="image/svg+xml"></object></td></tr><tr class="fig-caption"><td><b>(c) </b>Occluding patch size 20 × 20</td></tr><tr class="fig-image" id="d"><td><object data="https://static.hindawi.com/articles/mpe/volume-2015/239589/figures/239589.fig.005d.svgz" name="239589.fig.005d" type="image/svg+xml"></object></td></tr><tr class="fig-caption"><td><b>(d) </b>Occluding patch size 25 × 25</td></tr></table>

<div>Recognition accuracies versus different feature dimensions on ORL database with different occluding patch size.</div>

Mathematical Problems in Engineering

fig5

Figure 5

Figure 5: Graph Regularized Nonnegative Matrix Factorization with Sparse Coding 