New Tools for Watercolumn Feature Detection, Extraction, and Analysis - July 22nd & 23rd, 2014
Full Webinar Title: New Tools for Watercolumn Feature Detection, Extraction, and Analysis
Presentation Date(s): July 22, 2014 and July 23, 2014
Presenter(s): Mark Paton, Graphics and 3D Interactions Specialist, QPS
This webinar covered:
- Discussion of the open plugin infrastructure that makes creation of custom tools in the Fledermaus suite easy and attainable to users.
- Background of the Feature Detection tool.
- How to use the Feature Detection Tool and explanation of the tool options.
- Explanation of the Point Clusters SD object, a new Fledermaus SD object for analyzing potential seeps and other appropriate scientific data.
- Discussion of our plans to further automate seep detection and analysis.
Webinar prerequisites:
- Familiarity with the Fledermaus Suite, including Fledermaus module and FMMidwater module. File types, module functionality, etc. not fully explained.
Webinar Recording
Attendee Questions & Answers
Question | Answer |
|---|---|
| Can one export extracted features only line-by-line or can one process and export the features for multiple lines/GWC files in one go? | Current system is line-by-line, however we are looking at the best ways to batch across multiple lines/files for future work. |
| Which of all these tools work for are working MB sonar data that is not Kongsberg? | The core FMMidwater tool works for various sonars. The Feature Detection tool that is Mark is demonstrating and the ability to export Refracted Points is currently restricted to Kongsberg data. |
What is the plan for the characterization field? Is there a way to discriminate seeps from other water column artifacts? | The plan for the characterization field is to investigate algorithms for identifying between seeps, fish, and other water column artifacts. The first set of additional algorithms will be further work on using morphology of the clusters for the characterization and will be part of our fall release. |
Is there any potential to incorporate batch processing capacity into the plugin for processing multiple files and outputting results (e.g. lowest point coordinates) in a fully automated way? | Yes, this is something we are investigating for a future release. |
| What is the difference between a 3D point object for cluster analysis and a basic 3D point object? | The basic 3D Point object is our standard object for displaying points; you can export these points from FMMidwater and also from the Feature Detection tool directly, and you can create this object type by importing any supported point data format into the Fledermaus module using the Import Points option. The 3D point object for clustering looks the same visualization-wise, but has additional options in the Fledermaus module for analyzing and exporting the data. |
| Did you just say FMMidwater only works with Kongsberg data? | The FMMidwater tool works for various systems and data formats (FMMidwater Supported Data Formats). The Feature Detection tool (and the option to export Refracted Points) is currently restricted to Kongsberg data. Support for Reson and R2Sonic feature detection is being developed for an upcoming release – we are currently targeting fall 2014. |
| The feature contacts seem pretty focused on Seeps, how does it handle large detections such as wrecks or underwater structures and infrastructure? | We have not tested it much on these types of features, but would expect it to work well with anything that has a clear contact. Normalization/depeckling look for consistency in the signal and this should be present with structures. |
| Would you be able to recommend some threshold settings for EM2040 or 3002 data? | If you are in shallower water (EM2040), you'll have a higher rate of pinging and more beams, so you may want to increase the threshold, beams, samples, and success rate in the Despeckle Filter Options. Can't say for 3002 without seeing sample. You might want to play with it using the manual filter options first to get a feel for good setting values. |
| Will the "older" options – manual despeckle, filtering – work on SEG-Y data? | We think they will work on SEG-Y, but have not tested. |
A higher quality WMV recording of the webinar and a PDF of the presentation are also available for download. Please email webinars@qps.nl if you would like to obtain the download location for this webinar.