|
|
| |
|
|
|
Computational vision is at the heart of biomedicine and robotics, but is still quite primitive when compared with our own visual sense. We effortlessly demonstrate enormous flexibility and generality, which hides its staggering complexity: Nearly half of the primate brain processes visual information. Our group is attempting to put the requirements of vision systems together with insights from neurophysiology and the rigor of mathematics to develop an abstract theory of computational vision. Based on differential geometry, it leads to methods of curve detection and shading and texture analysis.
My group is working on early vision, grouping, and generic shape analysis. Recent projects and research topics are demonstrated below. For a recent overview paper on computing in cortical columns, a topic at the heart of our research, please click here.
To view additional images, drag cursor over links (click on links for full size images).
|
||
|---|---|---|
| Original Image | Detection Results | |
|
|
View:
Logical/Linear Edges Canny Logical/Linear Dark Lines Logical/Linear Bright Lines |
|
|||
|---|---|---|---|
| Noisy initial measurements | Relaxed Texture Flow | ||
|
View: |
|
|
View: (please hold) |
|
|
|---|---|
|
View: Shading flow field variance Shading flow field (zoom) Edge measurements Edge classification (white=fold, black=cut) |
|
|
|---|---|
|
View: Geometric Rectification Initial edge tangents Improved Space & Orientation quantization 3D Reconstruction |
|
|
|
|---|---|
|
View: Noisy Image Relaxation in action |
|
|
|
|---|---|
|
View: Failure of existing models to predict statistics Measured vs. Predicted statistics in the orientaiton domain |
|
| Bouton distribution image by Bosking et al. 1997 |
|
|
|---|---|
| Which shape is different? | Is the dot to the right of center? |
|
|
|
||
|---|---|---|
| Computed shocks | Corresponding shock graph | Matching via maximal subtree iosmorphism |
|
|
|