Assignment 12- Processing UAS Data in Pix4D

Introduction

In the final assignment of the semester, the class was introduced to Pix4D software.  A true orthomosaic of imagery will be created using this software with images from the Litchfield mine.  Before using this software and fully understanding it, it was required to go over the manual to get a basic understanding of the software.  While this software is relatively easy to use at first, there are many processes that one can do to advance their data.  In order to do so, you must learn about how to use the software and requirements your data must meet.  There were certain questions that needed to be answered which are seen below.  The link to the Pix4D manual which helped answer these questions can be found here.

• Look at Step 1 (before starting a project). What is the overlap needed for Pix4D to process imagery?
• High overlap, at least 75%, is required in order for high accuracy results.  This means that image acquisition needs to be carefully planned in order to obtain high end results.  An example of the minimum overlap needed for high accuracy results is seen below in Figure 1.

Figure 1: An example of the minimum amount of overlap needed for high accuracy results.

• What if the user is flying over sand/snow, or uniform fields?
• These areas have very little visual content because of the large uniform areas.  This means high overlap is needed in order to produce good results.  This overlap should be at least 85% frontal overlap and at least 70% side overlap.  The exposure settings should also be set in order to get as much contrast as possible in every image.
• What is Rapid Check?
• This is a way to check that the data output will turn out okay.  It reduces the resolution of the original images, lowers the accuracy and might lead to incomplete results.  However, it produces results much faster.  It is recommended to use in order to get a quick preview of the outputs to make sure everything look the way it is intended to.
• Can Pix4D process multiple flights? What does the pilot need to maintain if so?
• Yes, Pix4D can process multiple flights.  However, when designing the separate flights make sure that: 
• Each flight captures the images with enough overlap
• There's enough overlap between two flight plans (as shown in Figure 2)
• The different flights are taken under almost the same conditions

Figure 2: The amount of overlap needed to process multiple flights vs not enough overlap.

• Can Pix4D process oblique images? What type of data do you need if so?
• Yes it is possible for Pix4D to process oblique images.  These are images that are taken with the camera axis not perpendicular to the ground (as seen in Figure 3).  In order to produce good results, the oblique dataset will need 85% overlap and if possible, fly two rounds around the object.

Figure 3: Vertical vs. oblique images.

• Are GCPs necessary for Pix4D? When are they highly recommended?
• They are not necessary, but they are highly recommended when processing a project with no image geolocation.  If no GCPs are used, then the final results will not have a scale, orientation, or position information.  This means the images cannot be used for measurements, overlay, and comparison.  GCPs are highly recommended because they increase and verify the accuracy of a project.  
• What is the quality report?
• This is automatically displayed after each step of processing.  It checks the data to assure it has high quality.  It gives indicators of whether it is good or not throughout the long and detailed report.  A quick preview of the quality report example can be found by clicking here.

Methods

Now that some background information on Pix4D has been given, it is time to focus back on the Litchfield mine.  The images taken with the UAS were imported in Pix4D to begin the process.  It was recommended to create a smaller study area in order to make the image processing go fast.  However, for the final assignment it was decided to do the whole mine.  There are three options for processing options which are initial processing, point cloud and mesh, and DSM orthomosaic and index. Initial processing was only done first to make sure everything was working correctly. Once this was finished, the quality report was created.  As stated before, this gives an overview of the images and to what quality they turned out.  The images overall has high overlap except for on the edges of the image which wasn't a problem.  Once all of this looked good, point cloud and mesh and DSM orthomosaic and index were run.  These processes took over 10 minutes to run, however the results were well worth it.  A video of the finished results can be seen in the results section.

Results

The mosaic and DSM of the study are was imported into ArcMap and the results are seen below in Figure 4 and 5.  the DSM was also brought into ArcScene which is pictures below in Figure 6.  Pix4D produced highly accurate results which were very pleasing to see and were done with very little work (at least on the students end).  Flying and planning the UAS trip is a different story.

Figure 4: The DSM of the study area as seen in ArcMap.

Figure 5: The mosaiced image of the study area.

Figure 6: The 3D image of the study area as seen in ArcScene.

Conclusion

The results seen here are of very high quality which was because of a well planned UAS flight.  It is important when going out into the field to plan beforehand and to also have backup plans.  Things never really truly go as planned (as seen in this class and others).  However, having other plans and preparing before can still produce high end results.  Pix4D was a very useful software to learn and it would have been interesting to use more in the class if there was more time.  It is an easy to use software that has more advanced features that can be used as well.