Sunday, April 6, 2014

Developing Rule Sets Using eCognition

This video demonstrates the process of developing rule sets within eCognition.  Rule sets are knowledge-based expert systems in which, you, as the rule set developer, translate your knowledge into an automated workflow to extract information from the data.  This video assumes you are familiar with setting up workspaces and projects within eCognition.

Sunday, February 23, 2014

eCognition Workspaces & Projects

Prior to building rule sets in eCognition you need to create a project.  A project, much like an ArcMap document, references your data.  Storing your project within a workspace gives you additional control and functionality.  This video covers eCognition Workspace and Projects in detail.

Friday, February 14, 2014

Addison County LiDAR Point Density by Land Cover Class

After a bit of a delay due to the government shutdown the Addison County LiDAR is here!  I just finished processing the raster surface models, and after they are checked by the good folks at VCGI they will be made available to the general public

I thought it would be interesting to do a quick analysis of point density by land cover class.  For land cover data I used the NLCD 2006 product.  For each land cover class in Addison County I computed the total area of the land cover class, the mean LiDAR point density, and the standard deviation of the LiDAR point density.

Addison is a rural county, so not surprisingly the area totals are the greatest for the agricultural and forest classes.  Mean point density and the standard deviation for the point density is highest in the forest classes, which makes sense given the complex vertical structure of forests relative to the other classes.

This work was funded by a grant from AmericaView to the VermontView consortium.

Figure 1. NLCD 2006 for a portion of Addison County
Figure 2.  Point density, summarized using 5 meter grid cells for the same location as in Figure 1.  The presence of overlapping areas is clearly visible as they have higher point densities.
Figure 3. Point density statistics summarize by land cover class.

Saturday, February 1, 2014

Presentation Tips from Garr Reynolds

I am not the world's best presenter, but I know enough to not my my slides look like text from a scientific paper.  Before you present at the next conference do yourself and your audience a favor by spending 5 minutes reviewing Garr Reynolds Top 10 Presentation Design Tips.  Here's a hint - that table with 36 rows and 52 columns that you are going to show on the screen for 30 seconds does not belong.

Thursday, January 30, 2014

LiDAR Surface Models Graphic

I don't think this has enough cartoons to qualify as an infographic, but I hope it helps to illustrate the workflow used to create some common LiDAR raster surface model products from a point cloud.

Digital Elevation Model (DEM) - ground surface topography
Digital Surface Model (DEM) - elevation of all features including the ground, building, trees, etc.
Normalized Digital Surface Model (nDSM) - height of features relative to the ground.

Tuesday, January 28, 2014

Dark Object Subtraction

Dark object subtraction is a simple yet effective radiometric correction technique to remove atmospheric effects from imagery.  This video shows you how to perform a dark object subtraction using ENVI then examine your output using ArcGIS.

Sunday, January 26, 2014

Generating LiDAR Surface Models in ArcGIS

I never use ArcGIS to generate raster surface models from LiDAR point clouds, preferring tools such at Quick Terrain Modeler, SCALGO, or LAStools.  Nevertheless, people are obviously interested in an ArcGIS-centric workflow and this video shows you how.  In the video I take a classified point cloud and generate three raster surface models:
  1. Digital Elevation Model (DEM) - ground surface topography
  2. Digital Surface Model (DEM) - elevation of all features including the ground, building, trees, etc.
  3. Normalized Digital Surface Model (nDSM) - height of features relative to the ground.