|
|
Natural Image Segmentation |
|
Texture features in natural images are inherently heterogeneous -- different regions have distributions of different complexity and degeneracy. Lossy coding provides an ideal means of addressing this situation, leading to simple and effective algorithms for natural image segmentation. |
|
 |
Segmentation of gene expression data |
| Coming soon... |
|
|
|
Motion Segmentation |
|
Grouping features from multiple moving objects is a classical problem in computer vision. Our lossy coding approach is well suited for segmenting data lying on mixtures of subspaces, as naturally occurs in affine motion segmentation. Using techniques from sparse representation, we enhance our method to be robust to outlying, incomplete, and corrupted data. |
