TIPS

Clutter Removal

Aggregating your feature extraction results with the Feature Analyst® software means collecting or forming polygons into mass shapes that better represent your features of interest. The pre- or post-processing tool, Aggregate Features, informs the system to look for polygons with continuous pixel clusters with no fewer pixels than the number you designate. The system eliminates any “islands” and also the “holes” that are smaller than the prescribed number of pixels.

Pre-processing

By setting the aggregate area (minimum polygon size), you are defining the total number of pixels that comprise a feature. For instance, zooming in on a car, in the parking lot image shown below, one can see that the average car is made up of over 350 pixels. Setting the aggregate area to 350 pixels would filter out any polygons smaller than the size of the average car, but the 300-pixel cars would also be eliminated. So, to be safe, the aggregate area in this example has been set to 175 pixels. When aggregating features, we recommend that you start small and work your way up in the number of pixels.

Post-processing

Feature Analyst also provides aggregation as a post-processing tool. After you have already run your feature extraction pass, you can set the aggregate area larger than the clutter in your result layer. For instance, if you are extracting large buildings from a high resolution image, and each of the buildings is larger than 250 pixels, setting the aggregate area to 250 will filter out any small polygons that might otherwise come through, as well as any salt and pepper inside the building polygons. By setting the minimum polygon size larger than the clutter in your result layer but smaller than your target features, you can effectively eliminate much of the clutter from your feature extraction pass.

Click here to view the print version of this Tip of the Week.