Lab 2: Building Classification

Goals and Background

The main purpose of the lab is to develop skills in LP360 classifying building features. LP360 does not have the most robust building classification algorithms available on their software. The lab will have me using the building planar filter to classify building points. Additionally, the lab will have me performing manual clean up of incorrectly classified points. The manual clean up is the most time consuming operation in the process.

Methods

All of the following work was completed in LP360 utilizing the data from Lab 1 where I classified the ground and water points.

A backup copy of the data was created before starting this lab.

Building extraction with the use of a planar point filter

I created a new task in the Point Cloud Tasks. The task type was Macro and the task name was set to Lake Building Classification. Once the task was created I set the parameters as displayed per the guidance of my professor (Fig. 1). I had to establish the parameter for the maximum height. I set the points to display by elevation and utilized the Information tool in LP360 to identify the highest elevation in the study area (Fig. 2). I added an additional 10 feet to the dimension just in case there was a point or 2 which was higher which I didn't identify.

(Fig. 1) Building classification parameters set in the Point Cloud Task window.

(Fig. 2) Information window displaying the elevation of the highest building point.

After the parameters were set I tested the algorithm with the execute by polygon tool. The tool seemed to work for the most part but left portions of the building unclassified (Fig. 3).

(Fig. 3) Building roof tops left unclassified by the algorithm in LP360.

The tool parameters were adjusted a number of ways and nothing seemed to improve the results. So with the advice of my professor I ran the algorithm over the entire study area. I was left with many buildings which looked like the ones from Fig. 3. To correct the unclassifed points I created a basic filter to change the points from Unclassified to Building (Fig. 4). I utilized the Execute by Polygon tool and traced around the various roof tops to fill in the missing areas (Fig. 5).  I proceded to utilize the method through out the study area to fill in numerous roof tops which were left unclassified.

(Fig. 4) Basic filter created with in the Macro to change unclassified points to building.

(Fig. 5) Building points after using the basic filter to change the unclassified points to building points.

When filling in the unclassifed points I located a few of roofs which had been classified as ground which I missed in Lab 1. I utilized the filter from lab 1 to change the the ground points back to unclassified. Next I used my basic filter I created in this lab to change them to building points.

Results

(Fig. 6) Example of the results within my study area.

The results again were far from perfect and a good amount of clean-up is needed to make the classification acceptable for use. We will be focusing on the clean-up in Lab 4 when we perform QA/QC for our classification.


Sources

LAS, tile index, and metadata for Lake County are from Illinois Geospatial Data ClearingHouse. NAIP imagery is from United States Department of Agriculture Geospatial Data Gateway. Breaklines is from Chicago Metropolitan Agency for Planning.