IMAGE SEGMENTATION AND COMPUTER VISION
Active contour model without edges, also known as Chan-Vese model, detects objects in a given 2D or 3D image, and is based on techniques of curve or surface evolution, Mumford-Shah functional for segmentation, and level sets. Here, the model is used to segment 2D images and a 3D volumetric image. 3D animations are also shown.
Videos: 3D MRI (Explicit) |
3D MRI (Implicit) |
Transparent (Explicit) |
Transparent (Implicit) |
 |
 |
 |
 |
 |
 |
 |
 |
|
Lower Resolution: |
3D MRI - noise |
noise removed |
|
|
|
 |
 |
 |
|
Higher Resolution: |
|||
 |
 |
 |
 |
|
Images: Iso contours |
and Slices |
||
 |
 |
 |
 |
|
Animations: |
|||
|
|
 |
|
|
|
|
 |
 |
 |
 |
|
|
|
||
helicopter - Longbow |
plane - F22 |
America lights - Areas |
America lights - Cities |
 |
 |
 |
 |
MULTIPHASE MODEL
Multiphase model for Mumford-Shah image
segmentation, a generalization of the Active Contour without Edges model, allows to perform active contours, denoising, segmentation,
and edge detection. Multiple regions can be represented and detected using
two or more level set functions.
In medical imaging, the multiphase model can be used to
separate white/gray/black matter and to look for tissue loss in the brain:
|
Videos: |
Solid |
Transparent |
2 of 4
segmented |
|
Lower |
 |
 |
 |
| Higher Resolution |
 |
 |
|
Implicit - Dirac delta |
|
Implicit - Gradient |
|
 |
 |
 |
 |
Segmentation and Noise Removal
 |
 |
