Activity 12: Pix4D Mapper

Introduction

In order for the Pix4Dmapper software to process imagery it needs matched keypoints. Matched keypoints are two points from two different images that are in the same location. The more matched keypoints there are, the more accurately 3D points can be computed, so a high level of overlap is required for an accurate output. Recommended overlap is 75% frontal and 60% side. For area with little visual content; such as snow and sand, frontal overlap should be at least 85% and side overlap should be at least 70%. The exposure should also be set to get as much contrast as possible. Rapid check is a processing image that can be used to quickly produce low-resolution images to assess quality and completeness of the image while still onsite. Pix4Dmapper can process multiple flights however the pilot needs to maintain a similar altitude for the flights.   Oblique images can be processed, this data will need to be collected with the camera at a 45^o angle and succeeding images should be higher with a decreased angle. To combine oblique imagery with nadir images it is highly recommended to have GCPs and/or tie points, however GCPs are not necessary for Pix4D. After each step of processing a quality report will be displayed. The quality report is very a comprehensive report of how the process went, it includes a quality check, a preview of the images produced, number of images overlapped throughout the output, and much more.  This lab is designed to introduce these basic functions of the program using an image taken of the Litchfield mine, southwest of Eau Claire (figure 1). 

Figure 1. The study area

Methods

For this lab Pix4Dmapper Pro was used to create a 3D image of the Litchfield mine from a series of images collected using a Phantom UAV; to do this a new project was started in Pix4D and the images from the flight uploaded into the project. A quality report was automatically generated after the process was finished. The image was then used to experiment with a few of the various tools available in Pix4D mapper. A line measurement, a surface area, a volume calculation, and a video tour of the study area were all made. The image was also brought into Arc Scene to create another 3D rendering of the imagery.

Results/Conclusions

Pix4D mapper is a relatively user friendly program. The import and manipulation of imagery is rather straightforward, however the import process does provide a lot of information in the quality report that is a bit overwhelming. Measurements taken provided a reference of the overall size of the mine. Figure 2 displays the location of the measurements taken; the line length is 19.5 m, the surface area measured is 66.23 m², and the volume measured is 32.93 m ± 1.44 m³. Figure 3 is the 3D rendering of the study area made in Arc Scene, set with the image itself as the floating surface to improve the appearance. Finally a video was made to give a virtual tour of the image produced in PiX4D, attached below. 

Figure 2. imagery with location of measurements taken

Figure 3. Arc Scene rendering of imagery

 

Field Activity 11: Point Features Survey Using a Dual Frequency GPS

Introduction

This lab is designed to provide experience using a survey grade GPS, which is accurate within centimeters. The mapping grade GPS can result in error of up to a meter and some projects demand precision. For example, the mapping of a cemetery could not have error of that extent as overlap of burial plots could occur. For this lab the class surveyed a small section of the campus mall (figure 1) and mapped the results.

Figure 1. Reference map for survey area

Methods

To begin this lab the components of the survey GPS were explained. The GPS itself is mounted at the top of a tripod apparatus and there is a handheld component that is blue toothed to communicate with the GPS unit. The class paired up and each pair took two sets of points. One of the members positioned the tripod where the point was to be recorded by staking the front leg into the ground and then letting out the length of the other two legs as needed to place them into the ground as well. The tripod was then leveled using a leveler that is attached to the tripod at about waist height. When the tripod is in position the partner, who is holding the handheld device records the point (figure 2). This resulted in a total of 20 points collected in a random, stratified manner that was supposed to emphasize the area of relief in the study area. The information collected was then transferred into a text file used to create a point feature class in ArcMap. The point feature class was then used to create a series of continuous surfaces using the various interpolation methods used in Field Activity 5.

Figure 2. Data collection

Results/Discussion

In the sandbox survey performed in Field Activity 4, 434 points were collected in a square meter area. Each of the interpolation methods resulted in an accurate portrayal of the surface and the merits of each interpolation method could be evaluated based on these results. In this lab, only 20 points were collected in a much larger area. The interpolation methods; IDW, kriging, and spline employ algorithms not recommended for random sampling schemes as they will result in over representation of areas more densely populated with survey points. The TIN interpolation method is not recommended for areas away from survey points. While none of the interpolations employed were not advisable for the sampling scheme used, they should have resulted in a surface representative of the actual surface. Figure 3 depicts much of the survey area, which is a mound shaped area. When compared to figures 4, 5, 6, and 7; one can see that a lack of sufficient sample points resulted in inaccurate representations in all interpolation methods. Perhaps if the number of points had only been doubled to 40, an accurate continuous surface may have resulted. Furthermore, had it been included the natural neighbor interpolation method would most likely have been the best fit for a stratified sampling scheme. 

Figure 3. Surface of survey area

  Figure 5. TIN interpolation

Figure 4. IDW Interpolation

Figure 7. Spline interpolation

Figure 6. Kriging interpolation