Our team has established a basic set of characteristics used to analyze the products generated from Ecosynth. Parameters such as RGB color distribution among points or comparing point cloud densities can be assessed and used to create a "report card" that presents a summary of information we feel is important for examining data quality. An example of such a report card is shown to the right. All of the data analysis is done by running python scripts on different data outputs created during the Data Processing stage. A flow chart is presented on this page to act as a visual guide.
Data Analysis Components
Examining the distribution of RGB color among points can be done right after the Noise Removal Filter is run. For this analysis, the noise-removed point cloud file (.ply) serves as an input for the python script. The script will sort each color channel into bins and calculate the relative frequency of that bin. The result is a plotted graph (see the report-card for an example).
Examining the density of points in a given volume is another characteristic to analyze. Point density gives us some insights as to how various software packages (VSFM, Photoscan) handle SFM (Structure from Motion) operations. The python script Point_Density_Histo.py is used to generate a histogram that shows the distribution of point densities for the data, as well as calculating the mean, max, and standard deviation. The data input used for this script can be from two sources:
- Density Raster converted to an ASCII File in ArcGIS
- Computed Grid Stats.csv
Examining the CHM difference map is a critical component for our analysis. The python script CHM_Deviation.py will calculate the distribution of difference, as well as calculating the mean, max, and standard deviation. This is particularly important because it is a direct comparison between Ecosynth and LiDAR (which for our purposes is our gold standard). By comparing our Ecosynth CHM's against LiDAR CHM's, we can see how well our process matches LIDAR.