03/14/17
Digital Surface Modeling Using Pix4D
This lab/post was different than some of the previous technical labs and posts. Compared to the technical format that we usually use this post will be broken into three parts. First, background on Pix4D will be provided with an emphasis on its capabilities. Second, the Pix4D software will be discussed. Finally, several maps will shown that were the products of Pix4D software.
Part 1: Becoming familiar with Pix4D. How is the program used, and how is the data processed?
Pix4D essentially generates a three dimensional images. First, a drone will fly over the area of interest and take aerial photographs in a specific manner. Then Pix4D will overlap the photos pixels with specific ground points. The camera position is then calculated using an algorithm so a 3D image can be developed.
Using the Pix4D manual, some key questions can be answered regarding the dynamics of Pix4D.
1. What is the overlap required for Pix4D to process imagery?
A high amount of overlap is required to obtain accurate results. This means that a very specific plan to acquire the images must be put in place to insure the proper amount of overlap. When making the "image acquisition plan" the ground sampling distance must be known along with the terrain type and other project specifications. Failure to have a proper plan will result in low quality data.
2. What if the user is flying over sand, snow, or a uniform field?
The manual states the recommended overlap should have at least 75% overlap. There must also be 60% side overlap between flying paths. A uniform surface such as those listed above will require at least 85% overlap and 70% side overlap.
3. What is rapid check?
Rapid check quickly determines if the images taken are good enough to sufficiently cover the area of interest. It determines whether the image can be processed.
4. Can Pix4D process multiple flights? What does the pilot need to maintain if so?
Yes, Pix4D can process multiple flights. However, a specific number of overlap points are required (figure 1). Figure one displays what this might look like.
Figure 1.
5. Can Pix4D process oblique images, and what type of data would you need to do so?
Pix4D can process oblique images, but the camera must take pictures at a 90 degree angle to the ground or a 45 degree angle to the ground.
6. Are GCPs (ground control points) required for Pix4D or are they just recommended?
GCPs are not required, however, using GCPs will greatly increase the accuracy of the data. The project can then be placed on the exact position of the Earth.
7. What is a quality report?
After the points taken by the drone are processed a quality report is then generated. The quality report will include information pertaining to the processing of the image. If there are questions about the integrity of the image generated the quality report should be viewed.
Part 2: Using Pix4D.
After starting Pix4D one will simply go to "start new project". Next, the images must be added from the drone flight. This is done by copying all of the images from the drone flight into Pix4D.
Below are the image points as shown in Pix4D (figure 2).
Figure 2.
Then, you must select the type of project you with to create. We selected create a new 3D map (figure 3).
Figure 3.
After the data is uploaded it must be processed (figure 4). We first selected initial processing making sure the other options were unchecked. It is important to make sure that "Point Cloud and Mesh", and "DSM, orthomosaic, and Index" are unchecked initially. After the initially processing the two previous options were then checked and ran.
Figure 4.
Once processing is complete a quality report is then generated (figure 5). Figure five was taken directly from the quality report.
Figure 5. Fist displayed in the quality report is the initial quality check. Our initial quality check was very promising. A total of 68 images were used. 68 out of 68 images calibrated correctly.
This image taken from the quality report shows that our image had a great amount of overlap as indicated in green. There were a couple small areas with poor overlap on the outsides of the image. This is because there were no other images taken outside the AOI.
Here are some of the geolocation details.
Next, an animated fly over was created to provide a great view of the mine.
After the animation was created I wanted to take a volume measurement from one of the sand piles (figure 6). This was done for future reference. The volume measurement was fairly straight forward. It simply required digitizing the desired area. As shown below.
Figure 6. The total volume of this sand pile was 7195.05 meters cubed. The error was + or - 76.10 meters cubed.
Part 3. Maps
The first map created was a map showing the 2D aspect of the Litchfield Mine (figure 7). This was done in ArcScene and ArcMap. The DSM (digital surface model) portion displayed on the top of figure two clearly shows the physical features of the mine. The individual piles are clearly shown whereas the mosaic on the bottom is less clear.
Figure 7. Here is the Litchfield Mine shown in 2D. The sand pile that had the volume measurement taken is labeled in the bottom portion of the map. Metadata is provided in the center of the map.
The second map created displays the Litchfield Mine in 3D (figure 8). In ArcScene the base heights were set to 1 to best display the 3D aspect of the mine. In 3D the mosaic image on the bottom of figure 8 better portrays the mine compared to the 2D mosaic image. An advantage of the mosaic image is that equipment can be portrayed whereas in the DSM equipment cannot be made out.
It should also be pointed out that the pile that the volume measurement was taken from is shown as the dark pile in the mosaic in the west central portion of the map.
Figure 8. The Litchfield Mine shown in 3D. The same metadata used in figure 7 applies to figure 8.
Conclusion:
Pix4D turned out to be a great tool for processing 3D images. There are also some features that are very practical such as the volume measurement feature and the animation feature. The manual proved to be very informative and easy to use. The main critique would include the time required to initially process the images. This can take a significant amount of time.
Sources:
No comments:
Post a Comment