Table 1.
Steps used to identify objects in confocal images of the corneal stroma
Step | Process | Purpose |
---|---|---|
1 | Inhomogeneity correction | Adjust local brightness to make brightness uniform across field. |
2 | Sigma filter | Reduce high-frequency noise. |
3 | Adaptive histogram equalization and low-pass filter | Redistribute intensities in image uniformly across full intensity range. |
4 | Threshold, 70% full range | Convert to binary image. |
5 | Mathematical morphology | Remove very small objects (noise), smooth edges of large objects, open bridges, and close gaps. |
6 | Subdivide each object | Separate cells that overlap (Table 2). |
7 | Sort objects by size | Objects can be addressed individually. Each object can be used as a mask to the original or processed images. |