Ticket #450 (closed Task: fixed)
Implement thinning step in Cayula-Cornillon front detection tool
| Reported by: | jjr8 | Owned by: | jjr8 |
|---|---|---|---|
| Priority: | Medium | Milestone: | 0.8 |
| Component: | Tools - Oceanographic - Fronts | Version: | |
| Keywords: | Cc: |
Description
MGET's current implementation of the Cayula-Cornillon front detection tool does not implement a step that thins the resulting fronts. This is not only due to the work required but also because thinning was only really important for the multi-image edge detection (MIED) version of the original algorithm, not the single-image edge detection (SIED) version that MGET implements. In the MIED algorithm, fronts from several images would be overlapped, producing a composite that needed to be thinned.
MGET's SIED implementation nonethless would benefit from a thinning step. Unlike the original SIED algorithm, MGET's implementation allows the user to configure the histogram window size. This can improve detection of certain fronts, particularly near areas of no data (e.g. land, clouds). But it also produces very thick fronts, making it difficult to use without running a separate post processing step to thin the fronts (e.g. the ArcGIS Spatial Analyst Thin tool). By implementing a thinning step directly within the tool, we can eliminate the need for that post processing and probably provide much better performance as well.
