Chandrajit Bajaj
 Introduction | Visualization Tool | Filtering | Gradient Vector Diffusion | Segmentation | Skeletonization | References

5. 3D Segmentation

Our segmentation algorithm is based on the fast marching method and includes the following steps: First, we compute and diffuse the gradient vector field of the original density map. Then the critical points (local maxima and local minima) are calculated and used as the seeds of the marching contours. The critical points, and accordingly the marching contours, are classified into several groups. Two marching contours merge into a single one if they come from the same group. Otherwise, they automatically stop on their common boundaries.


Fig. 3 Segmentation of densely grouped extracellular fibrillar proteins


Fig. 4 Segmentation of extracellular filaments being secreted from frog saccular sensory epithelium supporting cells