Processing UAS Imagery

Objective and Background

The objective of this exercise is to process imagery using Pix4D, a premier software for generating point clouds. This lab builds off of the previous lab which covered the basics of Pix4D. For that exercise the data was already processed, however this lab will focus on how to process this data. This includes calibrating ground control points, creating an image mosaic and exporting the scene as a shapefile to be mapped in ArcMap. The processed data was obtained from a UAS flight at South Middle School in Eau Claire, Wisconsin by Dr. Joseph Hupy of the University of Wisconsin Eau Claire.

Figure 1
Aerial View of Study Area as seen on Google Maps

Methods:

The Pix4d software is relatively easy to use, however a few parameters must be set before the data may be processed. The first step is to create a new project. This can be done on the opening menu screen. 

Figure 2
Pix 4D opening Menu

 After a project was started, a series of images provided by Dr. Hupy were uploaded. A very important step here was to change the camera settings to linear rolling shutter, which is the particular way that the UAS captured this data. Some processing options that can be selected include the ability to generate shapefiles which can than be imported into ArcMap for mapping. For this exercise the option to create a 3D map was selected. The next step was to import ground control points, or known coordinates in the area of study, to improve the accuracy of our final product. The last step is to uncheck the options for point cloud/mesh and DSM/Orthomosaic on the bottom processing bar. The initial processing can then begin. 

Following the initial processing, a quality check report is generated. This shows any potential errors that could affect additional processing. The error in this quality report was corrected in the following step. During this step, the Raycloud editor was opened and the GCPs were calibrated to match with their actual locations, marked by pink squares in the study area. Three calibrations were done for each GCP

Figure 3
Matching GCP in the software (small blue circle under green x) with real world location (pink/black square)

After the GCPs are calculated, the re-optimize option in the processing bar must be used to adjust the images for the corrected GCPs. The second two processing boxes were checked (the first is unchecked to save time) and after some time the processing finished. The final product was then imported into ArcMap to create a final map.

Results

Figure 4
Orthomosaic Map created from processed UAS Data.

Figure 5
DSM map created with processed UAS data

For this exercise two maps were created. The first of these was a map of the Orthomosaic generated from the data that was processed in Pix4D. This map also includes a locator map to show where within the state the study took place. Overall the final product turned out well. there is enough overlap between all of the photos to create a good mosaic. When compared to an actual satellite image, the difference in quality can be observed, however with the orthomosaic overall the quality is good.

Figure 6
Othomosaic (Top) compared to Satellite image (Bottom)

The second map was a DSM map that showed the surface elevation of the study area. A hillshade generated using the DSM which was created during the image processing. This was overlaid on an imagery basemap which was made 40% transparent so it can provide spatial reference but not distract from the DSM. The DSM was given a red to blue diverging color ramp. Areas  of high elevation on this map can be somewhat deceiving because many of those areas are trees.

Overall the Pix4D software is easy to use, however it should be noted that for this lab a very small UAS data set was used. Even with this small size processing took roughly twenty minutes. With larger data sets the processing may take several hours. 

Conclusion

UAS data collection and image processing is becoming a very high demand geospatial field. These last two exercises have provided the basic information on how to use Pix4D, a UAS imagery processing software, to generate point clouds, DSMs and Orthomasics. These were then imported into shapefiles and mapped with ArcMap software. This lab demonstrated only a few of the several applications of UAS data