Pix4D

 Introduction

We were introduced to Pix4D over the span of 2 weeks. The first part was to get comfortable with the tutorials and help information and the second week was about the data processing. The software can be pretty demanding when it comes to processing and required us to use the computers in lab to complete the assignment. Pix4D has 4 steps it goes through when processing the data, Initial processing, analyzing the quality report, point cloud and mesh, and DSM.

Processing

The software looks for GPS points and starts arranging the photos accordingly. If any issues arise then it will show up in the quality report. In the quality report, the software will point out any issues with the data before it continues with the processing. Next, it goes through and starts meshing the images together. Finally, the software gets to the final stage where it has all the images put together and ready for viewing. Figure 1 here is an example photo from one of the processes. This is what the final product looks like.

Figure 1: A processed Raster

Final point clouds

Depending on how many images are taken will depend on how clear the processed photo will come out. Figure 2 is of a Raster that only had 15 photos. This made the picture have a lot of black space where the resolution wasn't good enough for the program to make an exact copy of the area. Next, there was a raster that included photos taken on angles. This didn't impact much of the processing and also made the image have "holes" within it. This can be seen in Figure 3. Finally, there was a Thermal Raster that we had to process. Due to the file type, it cannot show in Pix4D but after we transfer the file into ArcGIS the image can be seen. The Raster that was from this is Figure 4, The triangles at the top are where the photos were taken but again due to the file type, there is no image below. 

Figure 2: A Raster with only a few photos

Figure 3: A Raster with photos taken not at 90 degrees

Figure 4: The invisible thermal raster

GCPs

We then looked into using ground control points to give more accurate positing data for the images. This can be useful if the images aren't geotagged with an onboard GPS. After inserting the GCPs into the data you need to reoptimize the point cloud in order to take into account the GCPs. Once they are imported you must tell Pix4d where the GCP is on 4 images, this gives it some reference points to figure out exactly where the GCP is in relation to the images taken. Figures 5,6,7 are of 3 different aircraft and with 5 GCP points. There were 6 GCP points available but due to the lack of images in some of the data, it could cause the mapping to be messed with if it was included so there are only 5 GCP points. After the project is done being reoptimized you then should re-check the quality report to ensure that the geofencing box has a green check, ensuring that the program knows where the points are. Figure 8 is of this table and because it knows where the points were to a close degree.

Figure 5: A6000 images with GCPs

Figure 6: Mavic 2 Pro images with GCP

Figure 7: XT2 images with GCP

Figure 8: Green Checks within the report

Conclusion

By using Pix4D one can create very accurate 3D models that can be even used for measuring specific parts. The one drawback of this software is the hardware requirements and if your computer is on the low end you could be processing the images for a very long time.