The FME Oven – Never Too Many Ingredients Kurt Hartman Director of Technology, Accurate Assessment Group Ltd.
Jun 14, 2015
The FME Oven – Never Too Many Ingredients
Kurt HartmanDirector of Technology, Accurate Assessment Group Ltd.
Agenda
Introduction to Accurate Assessment Group Case Study – Video Logging Case Study – ERCB Data Questions
Introduction – Client Map
Introduction – Webmap Clients
Urban ClientsCity of Wetaskiwin
Town of Barrhead
*Town of Beaverlodge
*Town of Edson
*Town of Fox Creek
Town of High Level
Town of Peace River
Town of Redwater
*Town of Sexsmith
Town of Stettler
*Town of Two Hills
*Town of Wembley
*Town of Valleyview
*Village of Derwent
*Village of Myrnam
*Village of Willingdon
* Regional Sites
Rural ClientsBrazeau County
County of Athabasca
Camrose County
*County of Grande Prairie No. 1
County of Minburn
County of St. Paul
County of Stettler
*County of Two Hills No. 21
County of Wetaskiwin
Kneehill County
Lamont County
*Municipal District of Greenview No. 16
Municipal District of Opportunity No. 17
Rural Municipality of Wood Buffalo
Smoky Lake County
Westlock County
Wheatland County
Woodlands County
*Yellowhead County
Introduction – Municipal Information Integration
Video Logging
Video logging: a method of displaying video data within a GIS
Video is captured using a vehicle equipped with digital video cameras, precision GPS and on-board computers
Video Logging
Precise digital images are captured at regular intervals from GPS-equipped vehicles traveling at regular road speeds up to 100 km/ hr.
Video is post-processed using asset extraction/ identification software.
Assets can be located 80-100 meters from the vehicle.
Video Logging – Ingredients...
5,100,000 images 1,700,000 points
X,Y,Z GPS heading Image name Clip name Date/Time
Road network
Video Logging – Out of the oven...
4,800,000 images 1,600,000 linear referenced events
Linear reference values/keys Direction of travel Image name Image path
Batch files to create necessary directories Batch files to rename and move image files
Video Logging – Challenge 1
Attaching points to the correct road
Video Logging – Solution 1
Use Labeller to determine the orientation of the road in the vicinity of the point
Compare road orientation to GPS heading using Expression Evaluator
Video Logging – Solution 1 (Cont’d)
Using a Tester, determine if the difference between the road and heading azimuths are acceptable
Depending on the results from the Tester, assign the direction of travel for the point
Video Logging – Solution 1 (Overview)
Video Logging – Challenge 2
More than one pass on the same road
Blue and black points are going the same direction
Video Logging – Solution 2
Use StatisticsCalculator to determine for each video clip/road combination: Smallest linear reference value Largest linear reference value Total number of points
Video Logging – Solution 2 (Cont’d)
Use ExpressionEvaluator to determine the coverage that each clip has per road
Video Logging – Solution 2 (Cont’d)
Use a series of 3 Testers to validate which records should be included in the final dataset:
Test 1: If the point is part of the only video clip on that
road and it covers more than 10% of the road If it passes, include it If it fails, forward it on to Test 2
Video Logging – Solution 2 (Cont’d)
Test 2 If the point is part of a clip that covers more
than 25% of the road and the total coverage on the road is less than 110% This would handle scenarios where more than one clip
is needed to cover a road If it passes, include it If it fails, forward it on to Test 3
Video Logging – Solution 2 (Cont’d)
Test 3 If the point is part of a clip that covers more
than 75% of the road If it reaches this test, then it is likely a road that has
more than one pass To determine which of the passes gets included we
include additional variables Largest amount of coverage Most images Most recent date
If it fails, forward it on to the Unused feature
Video Logging – Solution 2 (Overview)
Video Logging – Challenge 3
Around 5,000,000 images (about 1.4 TB) requires intelligent file management
Developed a file structure that takes into account: Year of image Road name Alberta Township Survey township identifier Which camera (front, side, rear)
Video Logging – Solution 3
Points that are to be included in the final dataset are also forwarded to the Create Batch File process
Using a series of Testers, Concatenators and StringReplacers the Create Batch File: Create batch files that make the necessary
directory structure Creates batch files that move and rename the
image files Pushes the new image name and path back
into the final dataset
Video Logging – Solution 3 (Overview)
Video Logging – Final Translation
Video Logging – Finished Product
Video Logging – Benefits
Predictable result Reproducible result 36 person hours to create translation 1 person hour to run and validate 170 steps completed with 1 mouse click
ERCB Data
ERCB – Energy Resources Conservation Board Maintains Oil & Gas data for Alberta
Wells Pipelines Facilities
ERCB Data – Ingredients…
2 shape files 9 text files Cryptic field names Uses a lot of codes and abbreviations
Eg: Pipeline material type = “G”
ERCB Data – Out of the Oven…
4 feature classes Meaningful field names User-friendly data structure Replace codes and abbreviations with
“English” descriptions Eg: Pipeline material type = “Composite”
ERCB Data
Transformers used (118 in total): Joiner StringConcatenator SubstringExtractor FeatureMerger AttributeValueMapper PointConnector AttributeRenamer Tester
ERCB Data - Overview
ERCB Data - Benefits
Create user-friendly dataset Predictable result Reproducible result 118 steps completed with 1 mouse click
ERCB Data – Finished Product
Thank You!
Questions?
For more information: Kurt Hartman [email protected] Accurate Assessment Group Ltd. www.aag-gis.com